New Materials for Strontium Removal from Nuclear Waste Streams

Savvaki Neoklis Savva

Supervisor: Dr Joseph A. Hriljac

A thesis submitted to The University of Birmingham for

the degree of Doctor of Philosophy

The School of Chemistry

College of Engineering and Physical Sciences

University of Birmingham

September 2015

University of Birmingham Research Archive

e-theses repository

This unpublished thesis/dissertation is copyright of the author and/or third parties. The intellectual property rights of the author or third parties in respect of this work are as defined by The Copyright Designs and Patents Act 1988 or as modified by any successor legislation.

Any use made of information contained in this thesis/dissertation must be in accordance with that legislation and must be properly acknowledged. Further distribution or reproduction in any format is prohibited without the permission of the copyright holder.

Abstract

The primary aim of this project is to investigate potential new materials for application in ion exchange processes to remove 90Sr from nuclear waste streams. This work can be broadly split in to two sections, work on attempts to synthesise new materials and work to investigate ion exchange properties of two recently prepared materials AV-7, a synthetic analogue of tin- kostylvite and AV-3, a synthetic analogue of petarasite.

Synthesis on new materials was focused on metal silicate materials, in particular titanium, zirconium and tin silicates containing exchangeable group I and II cations. These synthesis attempts initially were focus on targeted mineral phases such as noonkanbahite,

BaKNaTi2(Si4O12)O2, followed by a series of brief surveys examining the effects of various changes to precursor gels such as concentration of bases such as NaOH, metal to silicon ratios and the presence of mineralizing agents such as sodium fluoride. The synthesis of two synthetic mineral phases potentially interesting for ion exchange is also reported here, titanite and fresnoite.

Ion exchange studies focused mainly on AV-7 and AV-3 but also included well known ion exchange materials for comparison such as clinoptilolite and Nb-doped crystalline silicotitanate and brief investigations in to the ion exchange of fresnoite and titanite. Ion exchange was followed using X-Ray fluorescence, ion chromatography and radioactive 85Sr exchanges measured using scintillation counters.

Acknowledgements

I would firstly like to thank my supervisor Dr Joe Hriljac for all his support throughout this

PhD project and all its difficulties. Thanks also to those at the NNL who helped with this work in particular my NNL supervisor Dr. Zoe Maher for some valuable guidance but also to

Dr. Scott Owns for useful discussions. Financial sponsorship from the School of Chemistry,

University of Birmingham and the NNL is also gratefully acknowledged.

A special thanks to past and present members of the Hriljac group, Tom Carey for help with my crystallography problems and George and Geoff for the time we spent in Boston. A special thanks to Evin who has been an incredible help throughout my time here.

I would also like to thank all those from the 5th floor of chemistry who helped in innumerable ways over the course of my work, from useful discussions over cups of tea to blowing of steam bouldering. A great thanks also to Dr Jackie Deans whose technical support has been a great help to me and everyone on floor five.

Finally I’d like to thank Katrina whose love, patience and support have helped with the hardest part of this work, the writing of it.

Contents

Chapter 1 – Introduction…………………………………………………………………....….1

1.1 Background ……………………………………………………………………………..1

1.2 Ion exchange ...... 3

1.3 Inorganic ion exchangers ...... 5

1.3-1 Natural inorganic ion exchangers ...... 5

1.3-2 Synthetic ion exchangers ...... 9

1.4 Potential materials for ion exchange – scope of this work ...... 16

Chapter 2 - Experimental……………………………………………………………………..17

2.1 Synthesis techniques ...... 22

2.1-1 Solid state synthesis ...... 22

2.1-2 Sol-gel synthesis ...... 22

2.1-3 Hydrothermal synthesis ...... 22

2.1-4 Microwave synthesis ...... 23

2.2 Characterisation techniques ...... 24

2.2-1 Fundamentals of crystallography and diffraction ...... 24

2.2-2 Neutron diffraction ...... 33

2.2-3 Rietveld refinement ...... 35

2.2-4 X-ray fluorescence spectrometry ...... 39 2.2-5 Scanning electron microscopy (SEM) ...... 44

2.2-6 Ion chromatography ...... 46

2.2-7 Radiotracer studies ...... 48

Chapter 3 - Titanosilicates and zirconosilicates………………………………………………45

3.1 Introduction ...... 51

3.2 Experimental ...... 53

3.2-1 Noonkanbahite & related phases synthesis gels ...... 53

3.2-2 Titanite synthesis gels ...... 55

3.2-3 Zirconosilicate synthesis gels ...... 56

3.2-4 Ion exchanges ...... 58

3.3 Results and discussion ...... 58

3.2-1 Noonkanbahite and related phases ...... 58

3.2-2 Titanite ...... 62

3.3-3 fresnoite ...... 68

3.3-4 Zirconosilicate synthesis ...... 72

3.3-4a Synthesis and characterisation of synthetic petarasite (AV-3) ...... 75

3.4 Conclusions ...... 94

Chapter 4 - Tin silicates ……………………………………………………………………...89

4.1 Introduction ...... 95

4.2 Experimental ...... 96

4.2-1 Hydrothermal SnO2-SiO2-NaOH synthesis survey ...... 96 4.2-2 Microwave TinII survey ...... 97

4.2-3 AV-7 - synthetic tin kostylvite ...... 98

4.2-4 Ion exchanges ...... 98

4.2-4a Bulk ion exchange ...... 98

4.2-4b Low concentration ion exchange ...... 99

4.2-4c Fused bead XRF ...... 99

4.3 Results and discussion ...... 100

4.3-1 SnO2-SiO2-NaOH synthesis survey ...... 100

4.3-2 Microwave TinII survey ...... 101

4.3-3 AV-7 - synthetic tin kostylvite ...... 109

4.3-4 Ion Exchanges ...... 109

4.3-5 Structural characterisation ...... 114

4.3-5a Time of flight neutron diffraction ...... 114

4.3-5a Temperature stability and variable temperature X-ray diffraction ...... 124

4.4 Conclusions ...... 129

5.1 Introduction ...... 130

Chapter 5 - Ion exchange……………………………………...…………………………….124

5.2 Experimental ...... 131

5.2-1 AV-7 Competitive strontium ion exchange ...... 131

5.2-2 Ion Chromatography ...... 131

5.2-3 Initial 85Sr exchanges ...... 132 5.2-4 Trace 85Sr ion exchanges pH 7 & 4 ...... 133

85 5.2-5 Trace Sr ion exchanges pH 10-11 ...... 135

5.3 Results and discussion ...... 136

5.3-1 AV-7 Competitive strontium ion exchange ...... 136

5.3-2 Ion Chromatography ...... 137

5.3-3 85Sr ion exchanges ...... 141

5.3-3a Initial exchanges...... 141

5.3-3b Trace 85Sr ion exchanges pH 7 & 4 ...... 144

5.3-3c Trace 85Sr ion exchanges pH 10-11 ...... 149

5.4 Conclusions ...... 158

Chapter 6 - Summary……………………………………………...... 151

6.1 Conclusions ...... 159

6.1.1 Titanosilicates and zirconosilicates ...... 159

6.1.2 Stannosilicates ...... 160

6.1.3 Ion exchange ...... 161

6.2 Further work ...... 164

List of Figures

Figure 1. 1 - The structure of clinoptilolite, viewed along c axis...... 8

Figure 1. 2 - Schematic diagram of SIXEP process ...... 8

Figure 1. 3 - Structure of crystalline silicotitanate, CST ...... 14

Figure 1. 4 - Na2ZrSi2O7·H2O viewed along [100]...... 17

Figure 1. 5 - Petarasite viewed along [001] ...... 18

Figure 1. 6 - AV-7 viewed along [100]...... 21

Figure 2. 1 - Diffraction through two rows of atoms...... 26

Figure 2. 2 - Reciprocal lattice over an Ewald sphere construction ...... 27

Figure 2. 3 - Debye Sherrer cones ...... 27

Figure 2. 4 - Reflection and transmission geometry for X-ray diffractometers ...... 33

Figure 2. 5 - Schematic diagram of Isis HRPD ...... 35

Figure 2. 6 - photoelectric effect resulting in fluorescence ...... 40

Figure 2. 7 - Penetration depth of electron beam and resultant emissions ...... 44

Figure 2. 9 - Schematic of Ion chromatography system ...... 46

Figure 2. 10 - Schematic of conductance cell ...... 47

Figure 2. 11 - Example of photoelectric effect generating photons due to gamma rays ...... 50

Figure 2. 12 - Schematic of Ge semiconductor photomultiplier detector ...... 50

Figure 3. 1 - Noonkanbahite along (100) ...... 52

Figure 3. 2 – XRD patterns of the first synthesis attempt of a new titanosilicate ...... 59

Figure 3. 3 - XRD patterns of sol-gel attempts at noonkanbahite ...... 60

Figure 3. 4 - XRD patterns of scherbakovite attempts ...... 61

Figure 3. 5 - XRD patterns of CaOH - TiO2-SiO2 system syntheses ...... 62

Figure 3. 6 - Rietveld refinement of Ca-titanite using TOPAS Academic 4.3 ...... 63

Figure 3. 7- Refined structure of Ca-titanite...... 64

Figure 3. 8 TGA trace of Ca-titanite between 30 and 1000 °C under N2 ...... 67

Figure 3. 9 - Rietveld refinement of fresnoite phase with excluded areas ...... 69

Figure 3. 10 - Refined structure of fresnoite viewed along (001) above and (010) below ...... 71

Figure 3. 11 - XRD patterns of the ZrO2-SiO2-NaOH-NaF system ...... 73

Figure 3. 12 - XRD patterns of synthesis attempts at Sr-zirconosilicate ...... 74

Figure 3. 13 - Rietveld refinement of as made petarasite (AV-3) ...... 76

Figure 3. 14 - Refined structure of a petarasite (AV-3) along (001)...... 76

Figure 3. 15- SEM image of AV-3 ...... 81

Figure 3. 16 SEM image of AV-3 g ...... 81

Figure 3. 17 EDX spectrum of SEM images indicating Sr/Si overlap ...... 82

Figure 3. 18 - Rietveld refinement of AV-3 ...... 83

Figure 3. 19 - Refined structure of Sr-exchanged AV-3 viewed along (001)...... 83

Figure 3. 20 - TGA traces of AV3 (above) and Sr-AV3 (below) including DTA trace...... 88

Figure 3. 21 - Variable temperature XRD patterns of AV-3 ...... 89

Figure 3. 22 - Rietveld refinement of parakeldyshite ...... 90

Figure 3. 23 - Refined structure of Sr-parakeldyshite...... 91

Figure 4. 1 - Typical PXRD pattern from NaOH-SnO2-SiO2 series ...... 101

Figure 4. 2 - Diffraction patterns of SnII:SnIV chloride microwave synthesis attempts ...... 102

Figure 4. 3 - Pawley refinement of new Sn-Si-O phase using Fm-3m space group ...... 103

Figure 5. 1 - Calibration curves for ion chromatography column…………………………..132

Figure 5. 2 - Material comparison for 100 ppm Sr ion exchange as measured by ion chromatography ...... 138

Figure 5. 3 - Example chromatogram for AV-7 before and after ion exchange...... 139

Figure 5. 4 - Example chromatogram for AV-3 before and after ion exchange...... 140

Figure 5. 5 - Batch distribution coefficients for competitive 100 ppm Sr ion exchange of AV-3 as measured by ion chromatography ...... 141

Figure 5. 6 - Batch distribution coefficients for initial 100 ppm 85Sr ion exchange experiments

...... 142

Figure 5. 7 - Structure of titanite illustrating pore diameter ...... 144

Figure 5. 8- Batch distributions for multiple materials ion exchanged with 100ppm Sr only at pH 7 ...... 146

Figure 5. 9 - Batch distributions for multiple materials ion exchanged with Sr only at pH 4 147

Figure 5. 10 - Batch distributions for multiple materials ion exchanged with Sr and competing cations at pH 7 ...... 148

Figure 5. 11 - Batch distributions for multiple materials ion exchanged with Sr and competing cations at pH 4 ...... 148

Figure 5. 12 - Comparison batch distribution coefficients for strontium of multiple materials exchanged in a range of different solutions ...... 150

Figure 5. 13 - Image of AV-7 post HIPing ...... 152

Figure 5. 14 - XRD pattern of HIPed AV-7 ...... 153

Figure 5. 15 Image of AV-3 post HIPing ...... 154

Figure 5. 16 - Refinement of HIPed AV-3 to identify parakeldyshite phase ...... 154

List of Tables

Table 3. 1 – Titanosilicate synthesis gel preparations ...... 53

Table 3. 2 - Titanosilicate synthesis gel attempts ...... 54

Table 3. 3 - Scherbakovite synthesis attempt gels ...... 55

Table 3. 4 - Titanite synthesis gels ...... 56

Table 3. 5 - Sr-Zr-Silicate synthesis attempts gel ...... 57

Table 3. 6 - Na-Zr-SiO2 -F synthesis gel preperations ...... 57

Table 3. 7 - Rietveld refinement parameters of Ca-titanite ...... 65

Table 3. 8 - Bond distances and angles from Ca-titanite Rietveld refinement ...... 66

Table 3. 9 - Rietveld refinement parameters and atom positions for Fresnoite r ...... 70

Table 3. 10 - Selected refined bond distances and angles of fresnoite ...... 72

Table 3. 11 - Rietveld refinement parameters of petarasite (AV-3) ...... 77

Table 3. 12 Selected bond distances from refinement of AV3 ...... 78

Table 3. 13 Selected bond angles from refinement of AV3 ...... 79

Table 3. 14 - Compositions of solids, reported as oxides as measured by XRF after initial exchange experiments...... 80

Table 3. 15 - Rietveld refinement parameters for AV-3...... 84

Table 3. 16 - Refined bond distances and valence sums for AV-3 ...... 85

Table 3. 17- Refined bond angles for AV-3 ...... 86

Table 3. 18 - Rietveld refinement details of parakeldyshite ...... 92

Table 3. 19 - Selected bond distances from the refined structure of parakeldyshite ...... 93

Table 3. 20 - Selected framework bond angles from the refined structure of parakeldyshite .. 93

Table 4. 1 - Mass compositions (g) of SiO2-SnO2-NaOH survey ...... 96

Table 4. 2 – Mass Compositions (g) for microwave SnII Survey ...... 97

Table 4. 3 - Compositions (g) for soidum SnII oxalate silicates synthesis gels ...... 98

Table 4. 4 - Compositions by percentage mass of solid samples derived from XRF ...... 110

Table 4. 5 - Integrated intensities and corresponding concentrations from ion chromatography

...... 113

Table 4. 6 - Refinement parameters for AV-7 ...... 116

Table 4. 7 - Refinement parameters for Sr-exchanged AV-7 ...... 117

Table 4. 8 - Refined bond distances and valence sums from neutron TOF diffraction of AV-7 and Sr-exchanged AV-7 ...... 119

Table 4. 9 - Refined bond angles from neutron TOF diffraction of AV-7 and Sr-exchanged

AV-7 ...... 120

Table 4. 10 - Refinement information for synchrotron data collection of heated AV-7 ...... 127

Table 4. 11 - Refined bond distances of heated AV-7 phase ...... 128

Table 5. 1 – Linear fits for ion chromatography calibrations 132

Table 5. 2 - Composition of ion exchange liquors - 85Sr concentration given in MBq due to a very low chemical concentration ...... 134

Table 5. 3 - Composition of ion exchange solutions A-D in ppm ...... 135

Table 5. 4 - Mole ratio compositions of solid AV-7 after ion exchange ...... 136

Table 5. 5 - Integrated intensities and concentrationof AV-3 Ion exchange liquor as measured by ion chromatography ...... 140

Table 5. 6 - Ion exchange solutions compositions in ppm ...... 150

Table 5. 7 - Leach testing water concentrations as measured by ICP-OES & ICP-MS ...... 155

Table 5. 8 - Normalized elemental mass loss (g m-2) ...... 156

Table 5. 9 – Composition of HIPed AV-7 and AV-3 as measured by XRF ...... 157

List of abbreviations

(P) XRD - (Powder) X-ray diffraction

BSE - Backscattered electrons

CST - Crystalline silicotitante

DTA - Differential thermal analysis

EARP - Enhanced actinide removal plant

EDS - Energy dispersive X-ray spectroscopy

GDF - Geological disposal facility

HDPE - High density polyethylene

HIP - Hot isotactic pressing

HLW - High level waste

HRPD - High resolution powder diffraction

ICDD - International centre for diffraction data

ICP- Inductively coupled plasma

ILW - Intermediate level waste

MS - Mass spectrometry

NNL - National nuclear laboratory

OES - Optical emission spectroscopy

PDF - Powder diffraction file

PTFE - Polytetrafluroethylene

RAL - Rutherford Appelton laboratory

SEM - Scanning electron microscopy

SIXEP - Sellafield ion exchange effluent plant

TEOS - Tetraethylorthosilicate

TGA - Thermal gravimetric analysis

TIP - Titanium isopropoxide

WDS- Wavelength dispersive X-ray spectroscopy

XRF - X-ray fluorescence

Chapter 1: Introduction

1.1 Background

With a growing number of nuclear power stations the storage and treatment of UK legacy waste is a growing concern. Radioactive waste is categorized according to the total radioactivity, heat generated and half-life as low level waste (LLW), intermediate level waste

(ILW) and high level waste (HLW). According to the 2013 radioactive waste inventory produced by the Nuclear Decommissioning Authority (NDA) and UK Department of Energy and Climate Change (DECC)1 there are 4.5 million cubic meters of waste in the UK. LLW is primarily composed of building materials and comprises the largest portion of this waste, approximately 94% by volume but less than 0.01% of the total radioactivity. The composition of ILWs can vary greatly but also contains building materials along with sludges, ion exchange resins and flocs. HLW is initially produced as a nitric acid solution containing fission products from the reprocessing of fuels. The long term storage goal of ILW and HLW is storage at a Geological Disposal Facility (GDF). However, in addition to solid wastes there is a large volume of contaminated water used for cooling and washing of contaminated materials; this presents a problem due to the difficulty in storing large volumes of radioactive material. It is this high-level waste which contains much of the strontium-90 (t½=28.79 years) and caesium-137 (t½=30.17 years) with which this thesis is primarily concerned; these are major contaminates in liquid effluent and are responsible for the bulk of heat generated in these wastes. Selective removal of strontium-90 and caesium-137 into a solid medium would result in a significantly more compact waste form which is of particular importance considering the space available at a GDF is a concern. Strontium-90 is of particular concern due to mimicking calcium in biological systems and is readily incorporated in-to bone tissue

1 where it will follow the decay pathway; a beta emission of 546 KeV to yttrium-90 and subsequently to zirconium-90 via a 2.28 MeV beta emission2. Caesium-137 can also be incorporated into biological systems where it will replace potassium and follows a decay pathway of beta decay to barium-137 followed by an X-ray emission of 662 keV.

In order to remove these radionuclides highly selective and robust ion exchangers are required as the waste solutions not only contain a wide range of cations which could potentially compete with the ion exchange targets, but also form a highly aggressive environment. In addition due to the high mobility of 90Sr and 137Cs a low leach rate is a vitally important property of an ion exchanger or a material it is transformed into as once stored any incidents such as flooding should not spill radioactive waste over a large area.

The ability for soils and natural minerals to absorb cations has been known and studied for a number of years3, however poor pH and chemical stability of these materials led to the development of organic ion exchange resins such as polystyrenedivinylbenzene4 which replaced many of the inorganic ion exchangers in use.

There are some processes in the nuclear industry, however, that have specific requirements for ion exchangers which these organic resins did not meet, most importantly good selectivity whilst maintaining resistance to ionizing radiation and high temperature. This lead to a number of inorganic ion exchange materials developed with use for the nuclear industry in mind5,6. Organic resins are common in other ion exchange applications, such as the petrochemicals industry but are not as commonly found in the nuclear industry7 where they act as filters for particulates such as cobalt and nickel and ion exchangers for ionic contaminants such as caesium and strontium.

Inorganic ion exchangers have been primarily used to treat waste solutions from decontamination operations and leaks in place of using evaporation type treatment, but despite a wide range of materials being tested few materials have demonstrated consistently good separation performance8–10.

1.2 Ion exchange

Ion exchange is a process whereby ions within a mobile phase are exchanged with loosely bound ions electrostatically bound to an insoluble stationary phase eg:

푆 − 퐻 + 푁푎+ ↔ 푆 − 푁푎 + 퐻+ (1.1) where S is a stationary phase such as organic resin or inorganic material.

The ion exchange process is heavily dependent on both kinetic and thermodynamic factors. In order for an ion exchange to be thermodynamically favourable it would be expected the ion to be removed from the mobile phase when compared to the ion in the stationary phase should:

 have a higher valence

 have a smaller solvated volume

 have a greater ability to polarize

 should precipitate less with any other reagents to form complexes

Kinetics of ion exchange are dependent upon the rate of particular transport pathways involved. A simplified form of the overall process is described as:

1. movement of ions from bulk solution to film or boundary layer surrounding ion

exchanger solid

2. transport through film/boundary layer to solid surface

3. transport through pores of solid to exchange site

4. exchange of ions

5. transport of exchanged ion outward through pores of solid

6. transport of exchanged ion outward through film/boundary layer

7. movement of exchanged ion into bulk solid

Typically the rate is controlled by steps 2 or 3 and these steps can be heavily affected by cation/pore size and charge distribution throughout the material.

When examining the performance of an ion exchange material there are 3 commonly used metrics:

 Cation exchange capacity – this represents the highest possible exchange often based

upon standardized measurement tests such as extraction with ammonium acetate, often

given in units of milliequivalents per gram.

 Batch distribution coefficient – The batch distribution coefficient (Kd) is an

equilibrium constant, used to measure of the uptake of an ion exchanger relative to the

ratio of exchanger to solution, it is given as:

(1.2) 푉(퐶0 − 퐶1) 퐾푑 = 푚퐶1

where:

V is the total volume of solution;

C0 is the initial concentration of the ion of interest;

C1 is the final concentration of the ion of interest;

m is the mass of ion exchange material used;

Whilst this metric may seem to be well suited to comparing different sets of ion exchange experiments, in practice they can vary significantly between experiments due to a number of factors affecting ion exchange and so it best to only use as a comparison within the same experimental conditions.

 Selectivity coefficient – an equilibrium constant used to determine how selective an

ion exchanger is for the desired cation when other competing cations are present:

(1.3) [퐶푠]푠표푙푖푑 푝ℎ푎푠푒[푁푎]푙푖푞푢푖푑 푝ℎ푎푠푒 푘퐶푠/푁푎 = [퐶푠]푙푖푞푢푖푑 푝ℎ푎푠푒[푁푎]푠표푙푖푑 푝ℎ푎푠푒

1.3 Inorganic ion exchangers

1.3-1 Natural inorganic ion exchangers

Clay materials such as kaolinite, bentonite and illite have been studied as ion exchangers extensively11,12, these materials are phyllosilicates, structures formed by parallel sheets of

5 silica tetrahedra, with two dimensional or layered structures comprised of Si /Al tetrahedra and Mg/Al octahedra, with cations such as Na+, K+, Ca2+ or Mg2+ to charge balance. Ion exchange sites are typically located on the surface, resulting in little to no ion sieve effect decreasing selectivity, however with good uptake and relatively low cost, the low selectivity can be of little importance. The mechanical properties of these materials, however, have reduced their use in an industrial setting as they are not suited to use in a flow through column-type exchanger due to poor fluid flow limiting their use to batch type reactors.

Mica type materials, such as biotite and muscovite have also been investigated for their ion exchange properties13. Micas are also phyllosilicates but have three dimensional structures composed of two silica tetrahedral layers corner sharing with an MO6 octahedral layer in between and two hydroxyl groups shared between three neighbouring octahedra. Common metals forming the octahedra are Fe2+ and Mg2+, but can be replaced by trivalent cation such as Fe3+ or Al3+, Fe3+ can also replace Si4+ in a tetrahedron, generating negative charge on the framework 14. The charge of the framework is balanced by the interlayer cations or anions which are exchangeable and result in the structure being expandable.

Zeolites are microporous (0.2-2 nm pore size) crystalline hydrated aluminosilicates defined by the general formula: (1.4)

푀2/푛푂퐴푙2푂3. 푥푆푖푂2 · 푦퐻2푂 where M is a non-framework cation.

The structure is built up of primarily Si and Al, but also can include Ga, Ge, Zn and Be, which are tetrahedrally coordinated to four oxygen atoms and linked to other tetrahedra forming secondary building units (SBU), such as single and double rings and cages such as

6 the β-cage which can be thought of as a truncated octahedron. The structural features of zeolites in turn form rigid 3-dimensional frameworks of pores and channels. AlO4 units are responsible for negative charge on the framework, counterbalanced by extra-framework cations or protons located in the pores or channels of the structure. The negative charge is located on an oxygen anion connected to the aluminium forming a Brønsted acid site if the site is a proton rather than a cation in the pore. The Si/Al ratio is a crucial feature of zeolites especially in terms of ion exchange, as this controls the overall framework charge and thus acidity and number of extra-framework cations. A lower Si/ratio and thus more negative framework would have a greater number of cations, but for the purpose of ion-exchange these cations may be too strongly bound for optimal exchange; counter to this too high a Si/Al ratio may lead to too few sites for ion exchange. The framework charge can also be modified by doping other atoms, such as Ga, into the framework to modify the charge15. According to

Löwenstein’s rule zeolites cannot contain Al-O-Al bonds within the framework; this prevents centres of negative charge being close to each other. The flexibility of the of zeolite type pore structure provides zeolites with their ability to selectively adsorb molecules based on size and shape. Whilst ion exchange properties are often very good, dissolution of Si in basic and Al in acidic conditions results in a narrow pH range for optimal performance16.

Figure 1. 1 - The structure of clinoptilolite, viewed along the c axis. Blue tetrahedra show SiO4 and AlO4 units. Purple spheres indicate cations and red the oxygen of water molecules.

Zeolites have been widely used as inorganic ion exchangers due to low cost and high

17,18 availability, particularly at the Sellafield Site Ion Exchange Effluent Plant (SIXEP) , where clinoptilolite, Figure 1.1, a natural mineral with the heulandite framework is used for this purpose as illustrated in a simplified diagram of the SIXEP process in Figure 1.2.

Figure 1. 2 - Schematic diagram of SIXEP process

However there are limitations to both the use of a natural zeolite. Zeolites tend to be most useful only within limited pH ranges, as not only can a high presence of protons affect ion exchange rates but extreme concentrations result in the dissolution of aluminium and silicon.

Thus it is necessary at SIXEP to first neutralize the effluent stream. With regards to using a

8 natural mineral, the composition can differ greatly depending on the site at which the mineral was acquired. The ion exchange properties of zeolites are highly dependent upon the Si/Al ratio and Al distribution throughout the structure as this affects the charge of the framework and the local cation environment. Thus reliance upon an inconsistent natural mineral is not ideal as the exact ion exchange properties cannot be fully known prior to testing and thus new synthetic materials for ion exchange have been a common area of research.

Elsewhere on the Sellafield site, the Enhanced Actinide Removal Plant (EARP) is responsible for the removal of beta emitting actinides. In contrast to the ion exchange processes of SIXEP, EARP uses the addition of sodium hydroxide to cause the flocculation of actinides by binding to them and forming insoluble metal hydroxides which sink and are easily separated from the effluent stream.

1.3-2 Synthetic ion exchangers

Unlike zeolites metallosilicates such as zirconosilicates, titanosilicates and stannosilicates do not follow Löwenstein’s Rule, which in conjunction with the metals usually forming octahedra as opposed to tetrahedra in the zeolites, allows for a much wider range of crystal structures. The pore size of these materials is usually on the lower side of the range of zeolites, with pore sizes usually in the range of 3-7 Å compared to zeolites which can vary from 3-20Å depending on the structure. Similarly to zeolites the ratio of Si/M is also of great importance with respect to ion exchange for the same reason of controlling the framework charge and number of cations. Metallosilicate frameworks can also be modified with the addition of other framework cations, in particular Nbv has been used to modify the framework charge of CST as discussed later in this section.

One of the most widely studied ion exchangers is crystalline α-zirconium phosphate

5,19–21 (Zr(HPO4)2 ·nH2O) . Zirconium phosphate has a layered structure with interlayer distances of 7.6A to 5.3A between nearest P-OH groups. Van der Waals forces between water molecules of the structure and P-OH groups are responsible for the interlayer binding. The protons are exchangeable from the structure, most selectively for Cs with a high capacity 6.63 meq/g 22.

Hexacyanoferrate, the pigment Prussian Blue, has been marketed as a medicine for heavy metal sequestration for some time, however it has also been shown to be incredibly selective for caesium when prepared in a granular form as the commercially available CsTreat® which uses potassium cobalt hexacyanoferrate. The FeII or FeIII forms can be prepared via precipitation with a metal salt solution, such as cobalt, to form the granular HCF particle with extremely good Cs selectivity which also has the mechanical properties desired by engineers to form columns. The high selectivity is due to cavities within HCF structure being very close match to size of caesium ion, allowing for an ion sieve effect to selectively ion exchange 23.

A series of papers by Dyer et al.24–28 focused on the exchange of MCM-41 (Mobil

Composition of Matter No. 41), a mesoporous silica material doped with heteroatoms Al, B and Zn. Unlike the previous materials which have been microporous with pore/channel diameters less than 2nm, MCM-41 is mesoporous with pores in the range of 2-50nm. Dyer used the incorporation of heteroatoms and different C-chain length to create a range of different pore sizes from 17.2 Å to 37.3 Å for a primary pore and to create a secondary pore of approximately 55 Å. Dyer used 0.05 g of solid in batch exchanges with a solid/solution ratio of 200 ml/g and nitrate salts of Li, K, Rb, Cs, NH4, Mg, Ca, Sr, Ba, and Al to obtain a total normality of 0.15. Isotherms of these exchanges indicated that for all MCM-41 ion

10 exchangers examined Cs+, Rb+,Na+ and K+ were preferentially ion exchanged over Sr2+ and other divalent cations; in fact this trend of monovalent cation selectivity over divalent cations is evident in the entire series of papers.

Sodium titanate (Na4Ti4O20·nH2O), commonly as monosodium titanate and sodium nonatitanate are poorly crystalline or amorphous layered materials which have been demonstrated to have good ion exchange properties for strontium particularly under basic condutions29. It can be prepared by using titanium dioxide powder in concentrated sodium hydroxide solution under reflux to obtain a poorly crystalline material. The exact crystal structure has yet to be reported in the literature but sodium titanates are generally built from edge-sharing TiO6 octahedra chains joined together through corner sharing to form layers with cations between them. Commercially available as SrTreat®, these materials have been implemented at some nuclear sites across the world 30,31, however, they have demonstrated a breakdown in performance in the presence of calcium, which is preferentially exchanged over strontium, drastically decreasing the use of sodium titanates in high salt waste streams. Poor exchange performance is also observed in low pH waste streams, thus calcium removal and neutralization of the waste stream is necessary in order to utilize these materials for strontium sequestration.

Potassium titanosilicate (K3H(TiO4)(SiO4)·4H2O), an analogue of the mineral pharmacosiderite, has been shown to have promise for strontium and caesium removal from waste streams under basic and neutral conditions32,33 . Doping of Ge into the framework for Si was shown to increase the uptake of both Cs+ and Sr2+ 34,35 although not by altering the framework charge but by altering the pore size of the material. A related material sodium titanosilicate (Na2Ti2O3(SiO4)·2H2O) is constructed of clusters of four corner-sharing TiO6

11 octahedra linked by SiO4 tetrahedra along two cell axes to form eight-ring pores of alternating

36 octahedra and tetrahedra joined together by corner sharing of the TiO6 octahedra . The sodium atoms are located both in the pores and in the framework between silicate groups, and some sodium atoms in the pores are replaced by protons due to limited pore volume changing the overall formula to Na1.64H0.36Ti2O3(SiO4)·2H2O. The tunnel-like pores are reported to be of optimal size for selective removal of caesium ions and the material has been shown to take up caesium over a broad pH range32,37.

A microporous titanosilicate ETS-10 ((Na,K)2Si5TiO13·nH2O) is another well-known ion exchange material with a three-dimensional structure composed of octahedral TiO6 units linked by bridging oxygen atoms to SiO4 tetrahedra. The pore structure contains wide channels approximately 8 Å in diameter38. Adsorption of heavy metals on to ETS-10 has been well studied over recent years39–43 including work by Pavel et al. on strontium and

44 2+ + caesium uptake . Pavel used 0.1 g of ETS-10 in 25ml solutions of Sr and Cs chlorides

n+ -1 from 4-9 meqM l for a contact ratio of 4000 mg/l. These solutions were kept at neutral pH and the exchange allowed to proceed for 24 hours. Pavel concluded that ETS-10 shows a slight preference for Cs+ uptake compared to Sr2+. Upon heating of the ion-exchanged ETS-10 it shown via SEM/EDX and XRD that whilst Cs remains homogenously distributed throughout the ETS-10 phase, Sr forms a separate fresnoite-like Sr2TiSi2O8 phase. Again this work highlights the different ion exchange behaviour of strontium in comparison to caesium.

45–48 Umbite, K2ZrSi3O9·H2O, is another well studied ion exchanger , which shares a similar pore structure to that of kostylvite (AV-7) one of the main foci of this work. Kostylvite both contain two separate channels, one six-ring of two pairs of corner-sharing tetrahedra joined by corner-sharing octahedra, and a larger eight-ring of alternating corner-sharing octahedra and

12 tetrahedra. Particularly interesting work by Fewox and Clearfield49 used time-resolved X-ray diffraction to probe the ion exchange process of a concentrated ion exchange solution of Cs+,

3.3 mM caesium in-situ into protonated umbite, H1.22K0.84ZrSi3O9·2H2O. The conclusions drawn from this work indicated that Cs+ exchange under these conditions is a two-step process; first the Cs+ exchange into one site followed by migration into a second more preferable site, which results in a structural change from monoclinic P21/c to orthorhombic

P212121. In-situ experiments such as this one, perhaps in conjunction with computer modelling, could potentially prove useful to further investigate the ion exchange mechanisms in these materials and contrast the behaviour of monovalent cations with divalent cations to improve the design of ion exchange materials.

Crystalline silicotitanate (CST, Figure 1.3) has been another heavily studied material, reports demonstrating its selectivity for Cs and Sr and highly crystalline synthesis lead to a great deal of interest in this and other metal-silicates 50,36. Dyer et al. used radiotracers to study the ion exchange properties of CST and a range of other titanosilicates51; ETS 10, ETS 4, AM 4 and

60 134 137 Na2Ti2SiO7·2H2O (Na-CST) were investigated with solutions of Co, Cs and Cs at both pH 7.1 and 5.9. Na-CST showed a significantly higher uptake of isotopes under all conditions. A general trend was observed where the performance of each of the titanosilicates improved with the slightly acidic solutions compared to the neutral solutions of isotopes.

Figure 1. 3 - Crystalline silicotitanate, CST where dark blue tetrahedra + are SiO4 units, cyan octahedra TiO6 units, yellow spheres Na cations and red spheres water molecules

It has previously been shown that substitution of Ti for Nb leads to an increase in the ion exchange capability52. Chitra et al. studied Nb-CST ion exchange for 90Sr and 137Cs 53,54 using simulated waste effluents to test the ion exchange ability in a realistic environment. The substitution of Nb for Ti resulted in a significant improvement in the ion exchange capacity for 137Cs. In addition to the doping of niobium into titanosilicates, germanium has also been doped into these materials simultaneously with niobium and shown to retain good, although not necessarily enhanced, exchange properties35. The uptake of 89Sr and 137Cs has further been shown to be inhibited by the presence of NaNO3, CaCl2, NaOH and HNO3; in the case of

33 CaCl2 the ion exchange capability is significantly decreased . Whilst CST shows a very high affinity for caesium when the concentrations of competing ions are low, this affinity drops off rapidly as the concentration of these increases past 0.5 mol-1 dm3, which limits the usefulness of this material unless either the material is modified to increase selectivity, or the waste stream is treated to remove calcium and sodium which is not practical. The addition of either

NaOH or HNO3 to alter the pH also results in a decrease in selectivity, however the drop off is not as rapid as with the inclusion of competing sodium or calcium. It is clear that CST, whilst

14 a good exchanger for caesium in pH neutral systems with low concentrations of competitive cations, is not as useful in highly basic/acidic or high sodium/calcium systems.

Strontium uptake was shown to be affected more by the presence of other cations. The presence of sodium ions in a pH neutral solution results in a decrease in strontium uptake and calcium ions result in an even more dramatic decrease. Whilst the uptake appears to be hindered by sodium ions at a neutral pH, the basic solution Kd values remain high, this could potentially be due to a change in the exchange mechanism with varying pH and this facet could warrant further investigation31.

A comparison of the ion exchange properties of clinoptilolite with a number of other ion exchangers, many of which are produced by UOP, has been made by Marinin and Brown 55; clinoptilolite, zeolite blend IONSIV® IE-96 (UOP) which consists of a mixture of natural chabazite and Linde A-51 zeolites, IONSIV® TIE-96 (UOP) a synthetic zeolite of undisclosed type modified with titanium, synthetic crystalline materials IONSIV® IE-

911(UOP) (a commercially available form of Nb-CST with an added zirconium hydroxide binding agent) and sodium titanate were tested along with a number of organic resins and composite materials. In the case of IONSIV® IE-911 and TIE-96 both demonstrate high distribution coefficients (30000 – 60000) but low selectivity when competing calcium cations are introduced, however, due to such a high uptake these materials can still be of use with regards to waste streams with low levels of competing cations. The zeolite materials exhibit a drastic Decrease in strontium uptake noticeable with an increase in calcium concentration, distribution coefficients drop from tens of thousands down to a few hundred; however, the selectivity remains high due to selective sorption occurring rather than ion exchange. The

15 composite and resin materials were noted to have significantly lower distribution coefficients, by 2 orders of magnitude, but superior selectivity and kinetic properties in most cases.

1.4 Potential materials for ion exchange – scope of this work

Whilst there are a number of different material types to study for ion exchange as previously discussed, the work reported in this thesis focussed on metal-silicate materials due to the flexibility of this system in terms of connectivity of polyhedra for structure and the possibility for isostructural substitutions, and the vast number of different structures possible as evident by the number of natural minerals based on silica tetrahedra with metal octahedra.

Natural minerals and the respective synthetic analogues have also been examined for ion exchange properties. As new mineral structures are discovered synthetic equivalents are reported and these materials are often investigated for a variety of possible applications. Of particular interest would be a natural mineral, noonkanbahite56 , which is related to the shcherbakovite and batisite materials with the formula BaKNaTi2Si4O14. This structure has yet to be synthesised or the ion exchange properties investigated. An interesting feature is a range of different size cation sites, which could potentially allow for selective uptake of a number of different cations on to the different sites.

The synthesis of zirconosilicates has been widely studied57–59 and a number of materials have been produced which could be potential ion exchange materials. The great number of these structures gives a wide variety of pore sizes determined by the motif of tetrahedra and octahedra in the pores. It is also worth noting that there are a number of materials isostructural

16 to zirconosilicate minerals with the zirconium switched for a different metal such as tin60. The size difference in metal octahedra could potentially allow for slight modifications to the pore size.

Na2ZrSi2O7·H2O, Figure 1.4, has been well characterized structurally, but little work has been performed to investigate ion exchange properties. Many metal silicates have shown good resistance to radiation damage and high pH stability, thus they could prove to be an interesting class of materials for investigation61,62. Furthermore recent work has shown that potentially desirable waste form phases can be achieved by thermal treatment converting to a new phase63.

+ Figure 1. 4 - - Na2ZrSi2O7·H2O viewed along [100]. Na sites are indicated by yellow spheres, and water sites by grey spheres. The pore diameter is approx. 4.7Å

Analogues to the natural minerals umbite, elpidite and petarasite64,65 offer possible materials which could potentially be useful for ion exchange. These materials have been structurally characterised and have been shown to have channels in the region of 2-6 Å diameter, with

17 both Na+ and K+ ions and shared sites which would suggest a possible affinity for Sr2+ as the ionic radius is between that of Na+ and K+.

Petarasite, Figure 1.5, Na5Zr2Si6O18(Cl,OH) ·2H2O, a synthetic form of which is termed AV-

3 has been recently prepared by Lin et al66 . AV-3 is of some interest due to the presence of

Cl- and HO- which could potentially result in a capability for anion exchange in addition to cation exchange.

Figure 1. 5 - Petarasite viewed along [001], showing the Cl- sites as green sphere, Na+ sites as yellow spheres and the water molecules are indicated by grey spheres. The diameter of the pore size is approx. 4.6 Å

There has been limited published work on the ion exchange properties of these materials, however, one such paper demonstrated how the ion exchange capability of the material

32 umbite, K2ZrSi3O9·H2O, depended greatly upon the pH of the system . A series of materials based on K2ZrSi3O9·H2O where framework Zr has been substituted for Ti in various amounts

(12. 25, 37, 50 and 75% Ti) were prepared along with protonated forms such as

32 K0.3H1.7TiSi3O9·2H2O and K0.5H1.5TiSi3O9·2H2O, by Clearfield et al. to investigate the

18 selectivity of these ion exchange materials for Li, Na, K, Rb and Cs under a range of pH conditions. The ion exchanges performed in this work took place over 10 days using concentrations of 0.05 M MCl-MOH (M=Li, Na, K, Rb, Cs) at room temperature. Clearfield found that doping Ti into the framework thus reducing the size of the channels resulted in a greater preference for potassium uptake from pH 2-6 whereas the more Zr-rich silicate exhibited a preference for rubidium and caesium cations. This work illustrates the link between pore/channel size and cation specificity and shows it to be an important factor when considering an ion exchange material and how to modify one to improve ion exchange properties. In the case of Cs+ it was reported that at pH 3.0 the uptake was 0.75 mequiv/g. which increased to a maximum of 2.0 mequiv/g. at pH 12. The uptake of potassium and sodium, however, are favoured over that of caesium over all pH ranges, with potassium being the most favoured. The uptake begins at a pH above 2, and increases to 1.3 mequiv/g. at pH 4 but increases suddenly after pH 6 where the uptake jumps to 4 mequiv/g. The uptake increases further as the solution becomes basic up to a maximum of 5.3 mequiv at pH 9 after which the uptake plateaus.

This research indicates how important the pH of the system is on the ion exchange. It should be noted that in the real world application of these materials in the nuclear industry, the pH of waste steams can vary widely and so good ion exchange properties over a wide range of pH is a very desirable property.

Stannosilicate materials have been less rigorously studied, with little to no work done on the synthesis of these materials. Of particular interest is a material termed AV-760 , with the ideal formula NaKSnSi3O9·H2O, Figure 1.6. The first synthesis was reported by Corcoran in 1992 before Rocha et al. reported the structure in 2000. Of some interest is the presence of shared

Na+ and K+ sites, which as previously mentioned, could indicate an affinity for Sr2+ exchange.

Furthermore, the structure reportedly has 4 Å channels in one dimension as opposed to fully enclosed cages within the structure, which would be more difficult for cations to access. Ion exchange studies have yet to be carried out on this material, so further investigation is warranted.

Other materials which have been reported, but have not as yet been investigated with ion exchange in mind include a synthetic tin equivalent of the natural mineral Penkivlksite-1M

(ideal formula Na2TiSi4O11·2H2O), AV-13 (Na2.26SnSi3O9Cl0.26·xH2O) and hafnium substituted equivalents 67–69. The substitution of hafnium into the structure could possibly be an interesting factor to investigate with respect to the effect upon ion exchange properties.

Hafnium being significantly larger than tin, could potentially open up the channels of the structure even further, also this material, as in AV-3, exhibits both anions and cations in the channels. This is another interesting factor, and whilst the group mentions forthcoming research on the possible ion exchange of nitrate anions into the structure no other anion exchange is mentioned.

Figure 1. 6 - AV-7 viewed along [100]. K+ sites are indicated by the purple spheres and the Na+ and K+ shared sites indicated by the yellow spheres. The pore diameter is approx. 5.1 Å

Chapter2: Experimental

2.1 Synthesis techniques

2.1-1 Solid state synthesis

The solid state synthetic approach is the most well used method for materials chemistry primarily due to its simplicity; reactants are ground together in a mortar and pestle before sintering at high temperature (often >1000°C). As this relies upon surface contact between reactants regrinding and heating is often necessary to ensure a phase pure product. Direct heating of solids often yields phases which are more condensed than those obtained via hydrothermal routes however, so this route is not often employed when seeking porous materials.

2.1-2 Sol-gel synthesis

The sol-gel method involves the preparation of a sol, a colloidal suspension of metal oxide particles in an alcohol solvent from a precursor, typically a metal alkoxide. The sol is then hydrolysed so as to form a gel which can then be aged to ensure completion of the reaction.

After ageing the gel can be dried to form a porous rigid network, which is then able to be heated at lower temperature compared to solid state reactions.

2.1-3 Hydrothermal synthesis

Hydrothermal and solvothermal synthesis are techniques where by liquids and gels are heated within a sealed chamber to generate pressure in order to crystallize materials from solution.

Typically water is the most common solvent used, however alcohols70 and ionic liquids71 can

22 also be used if the user is mindful about the pressure generated in the case of ethanol.

Pressure is usually generated autogenously, so that the saturated vapour pressure of the solvent and the level of filling determine the pressure at a given temperature.

The phases produced through hydrothermal techniques often differ, particularly in phase density, from those obtained via solid state routes primarily due to the presence of water/solvent but also due to the initial gelling process and the presence of any mineralizers such as F- ions which can also greatly influence the crystallization process.

Due to starting mixtures often being highly basic or otherwise corrosive, in the typical hydrothermal autoclave a resistant material such as PTFE is often used to contain the reagents, which also has the advantage of good mechanical properties up to 270°C and 150 bar. This vessel in turn typically resides within a steel body with a safety pressure release system in case of higher than expected pressure.

In the course of this work Parr (model 4749, 4744 and 4748) acid digestion vessels with

PTFE liners were used.

2.1-4 Microwave synthesis

Microwave heating is distinctive in that it produces even heating throughout a material unlike conventional convection heating ovens which rely on the thermal conductance of a material to heat the centre, which can often result in a large temperature gradient throughout a large object. Furthermore heating can be applied more directly to a material as only a polarizable material will be heated, so no energy is wasted heating air. This can enable very rapid

23 synthesis of some materials, as some zeolites have been shown to be formed by heating for as little time as 10 minutes compared to conventional heating which often takes days.

For this work a CEM MARS6 microwave was used along with EasyPrep acid digestion vessels which comprise a PTFE vessel charged with the reactant gel within a composite sleeve, secured closed with an HDPE screw mechanism. Various crystallization times were investigated over the course of the work ranging from 30 minutes to 8 hours, the temperature for all these reactions was set to 230°C controlled using an immersed temperature probe.

Pressure of the vessels could not be directly controlled or measured and so vessels were always filled to the same level (45 ml).

2.2 Characterization techniques

2.2-1 Fundaments of crystallography and diffraction

A crystal is a solid with a regularly repeating arrangement of atoms, the smallest representation of the repeating pattern is termed the unit cell. The unit cell is described by the lengths a, b, and c and angles α, β, and γ, which are referred to as the lattice constants. The relationships between the lattice parameters result in the 7 crystal systems, which when combined with the one of the lattice centrings give a total of 14 Bravais lattice types72:

 cubic (a=b=c, α=β =γ=90°)

 hexagonal (a=b

 orthorhombic (a≠b≠c, α=β =γ=90°)

 monoclinic(a≠b≠c, β ≠α =γ=90°)

 rhombohedral (α≠β ≠γ≠90°)

 tetragonal (a=b≠c, α=β =γ=90°)

 triclinic (a≠b≠c, α≠β ≠γ)

Combining the Bravais, lattice types with the 32 crystallographic point groups gives a total of

230 space groups, which all three dimensional unit cells must correspond to. The crystal can be described in terms of planes, described by the inverse of the fractional intercept (hkl) with the axis a, b, and c; so a plane bisecting halfway between the a axis and not intercepting either b or c would be the (200) plane.

Subsequent to the discovery of crystalline solids diffracting X-rays by Max von Laue in 1912,

William Lawrence and William Henry Bragg developed the first diffractometers based on the observation that when X-rays interact with atoms in a set of parallel planes separated by a distance d, the scattered X-rays can interfere constructively or destructively, giving rise to

Bragg peaks, according to the incident angle θ as per Braggs law73 shown in Equation 2.1 and

Figure 2.1;

푛휆 = 2푑ℎ푘푙 sin 휃 (2.1)

Figure 2. 1 - Diffraction through two rows of atoms74

When considering a single crystal it is simplest to examine the diffraction in terms of the reciprocal lattice and Ewald sphere; a sphere of radius 1/λ, where λ is the wavelength of the incident radiation on a crystal is termed the Ewald sphere. This represents the experimental conditions possible as a range of 2θ possibilities. The reciprocal lattice represents the scattering possibilities of the crystal. In the case of a 3D crystal the lattice would take the form of a series of points labelled with Miller indices corresponding to the planes at which diffraction will occur. Combining the Ewald sphere75 and reciprocal lattice, Figure 2.2, will indicate how diffraction will occur when a point on the reciprocal lattice intersects with the sphere. The diagram below depicts the case of a crystal orientated to cause diffraction on the

(–2,4,0) plane, the vector joining the origin to the reciprocal lattice point denoted, d*is the reciprocal lattice spacing which is equal to exactly 1/d.

Figure 2. 2 - Reciprocal lattice over an Ewald sphere construction74

In the case of powder diffraction, the powder can be considered a collection of randomly distributed crystallites in all possible orientations are represented each of which diffracts independently. This results in cones of diffracted X-rays, termed Debye-Scherrer cones,

Figure 2.3, which can be recorded by the detectors of diffractometers.

Figure 2. 3 - Debye Scherrer cones74

When the wave is diffracted by the plane of a crystal it can be described as a vector with an amplitude and phase with sine and cosine components. The intensity of the reflection from an ideal crystal with an ideal detector can be given by the sum of the scattering factors of the atoms on the plane with this phase component74,76.

2휋푖(ℎ푥+푘푦+푙푧) (2.2) 퐹(푆) = ∑ 푓푛푒 푛 where,

 F(S) is the observed intensity (structure factor)

 fn is the atomic scattering factor of the atoms in the plane of the unit cell

However, it is necessary to include a number of additions to this as the previous equation only considers an ideal crystal.

Multiplicity, j

As powder diffraction depends on the random distribution of crystallites, any set of Bragg planes with the same d-spacing will have a number of crystallites orientated for diffraction at

2θ, thus for reflections which have symmetry equivalent reflections (e.g. the (111) NaCl reflection has 7 other equivalent reflections) all of these equivalent reflections will contribute at the given angle.

Polarisation factor, P

Polarisation of the diffracted X-ray can change, the degree of which is dependent upon the initial polarisation of the X-ray being in plane or out of plane with the scattering. When the polarisation is perpendicular to the plane of scattering there is no change in intensity over 2θ, when in the plane of scattering, there is a drastic effect on the observed intensity over 2θ with the intensity being 0 at 2θ=90°. Unpolarized X-rays, as used by laboratory diffractometers, being a mixture of the two previous cases have a reduction centred on 2θ=90°. Use of a monochromator can also reduce the loss in intensity as the degree of polarization depends upon the diffraction angle of the monochromator, 2θM.

Polarization in plane : P = 1

Polarization perpendicular to plane: P = cos2 2휃

Unpolarized X-rays: P = (1+ cos2 2휃)/2

2 2 2 Unpolarized X-rays with a monochromator: 푃 = (1 + 푐표푠 2θ 푐표푠 2θ푀)/(1 + 푐표푠 2θ푀)

Lorentz factor, L

Whilst Bragg’s law implies that diffraction only occurs when the equation is satisfied exactly, in practice λ and d will have a degree of spread so that diffraction occurs over a range of angles centred on a mean, 2휃. This can be explained by considering reciprocal space; a point in the lattice remains on the Ewald sphere for a time period during the measurement process, these points are small but possess a size inversely proportional to that of the crystallite. When the detector scans though 2θ, the time the reciprocal lattice space spends in a diffracting

29 position on the Ewald sphere has an angular dependence; higher angle lattice vectors cut the

Ewald sphere close to a tangent and thus spend longer in a diffracting position increasing their intensity.

This results in a diffraction peak not being a delta function but being a curve with a height and width with an angular dependence:

퐿 = 푐/(푠푖푛휃 푠푖푛2휃) (2.3)

Where c represents a constant; in the case of diffraction the value of the constant is unimportant as only the relative intensities are of consequence.

Absorption, A

As X-rays pass through matter they are absorbed according to an exponential law:

−휇푡 (2.4) 퐼 = 퐼0푒

where I0 is the incident intensity, I is the reduced intensity after travelling distance t through a material with an absorption coefficient µ. The importance of this effect is dependent upon the geometry of the experiment conducted; in reflection geometry the absorption is unimportant as when all possible path lengths are taken into account the net absorption is constant with θ.

Temperature

The temperature affects the intensities in a more complex fashion as the effect changes from atom to atom in the structure. Modifying the atomic scattering factors to add isotropic displacement is a simple way to include this influence in the structure factor equation:

(2.5) 30

2 2 푓푇 = 푓 exp(−퐵푠푖푛 휃/휆 )

where T is temperature and B = 8π2, being the mean square displacement of an atom from its average position. These B factors, also termed Debye-Waller factors , increase with temperature.

The first four of these factors can be used as simple multiplicative corrections with an additional constant representing other influences such the sensitivity of the detector, so that:

2 퐼ℎ푘푙 = 푐푗푃퐿퐴 퐹ℎ푘푙 (2.6)

The temperature factor, however, must be included in the structure factor equation as a modified atomic scattering factor:

2 퐹(푆) = ∑ 푓푛푁푛 exp{2휋 푖 (ℎ푥 + 푘푦 + 푙푧)} exp( − 퐵푠푖푛 휃/휆) (2.7) 푛

Modern diffractometers generate X-rays beams through an X-ray tube, a sealed vacuum tube containing a metal filament cathode and a water cooled anode. By way of a large potential difference, electrons are accelerated towards a metal target resulting in the excitation of the electrons within the metal leaving vacancies to be filled by higher energy electrons. The relaxation of these electrons results in the emission of an X-ray and heat.

The transitions between electron shells of a target metal produce characteristic X-rays of that transition. For copper when an electron drops from the L shell to a K shell, the Kα1 emission has a wavelength of 1.5406 Å.

A continuum of background radiation termed Bremsstrahlung (braking radiation) is also observed due to the rapid deceleration of electrons when they are fired at a metal target and deflected by a charged particle. The decrease in the kinetic energy of the electron necessitates the emission of electromagnetic radiation in order to conserve energy.

Many diffractometers are fitted with monochromators either immediately after the X-ray tube and/or before the detector as seen in Figure 2.4. monochromators ensure a wavelength of as narrow a range as possible, a material where the absorption edge falls between the Kα and Kβ of the target material is used, so that it acts as a filter decreasing the intensity but ensuring monochromatic radiation is obtained. In lieu of a metal filter type monochromator, a single crystal, typically of germanium, silicon or quartz, set at a specific angle can be used to collect monochromatic radiation according to Bragg’s law72.

Figure 2. 4 - Reflection (top) and transmission (bottom) geometry for X-ray diffractometers52

2.2-2 Neutron diffraction

Whilst neutron-mass scattering is relatively weak in comparison to X-rays, neutron diffraction still proves to be a useful technique. The scattering of neutrons is based upon the nucleus of an atom, as opposed to the electrons in the case of X-rays. This allows for better sensing of lighter atoms, particularly when in the presence of heavier ones. Furthermore neighbouring

33 elements on the periodic table and even different isotopes of the same element often have significantly different scattering cross sections which allows for better differentiation.

However due to the weak scattering, large sample sizes are required and the signal-to-noise ratio is typically much weaker than in XRD.

Neutrons are typically generated in one of two ways: from a nuclear reactor or from a pulsed spallation source. Neutrons generated from reactors, termed ‘thermal neutrons’, typically have a wide range of energies and so a moderator material, such as water or graphite is necessary to slow the neutrons down in addition to a crystal monochromator to select the desired wavelength. In the case of a spallation source neutrons are generated by bombarding a metal target (e.g. Pb, Ta or Hg) with energetic particles such as protons accelerated up to 1 GeV.

ISIS at the Rutherford Appleton Laboratory, the facility used in this work, generates neutrons via spallation, using a particle accelerator to fire a pulsed beam of protons at 0.8 GeV at a tantalum target. The pulsed polychromatic nature of the neutron beam lends itself to collecting data via a time of flight (ToF) method rather than varying the energy of the beam or the angle of the detector. As the wavelength of a neutron is related to the energy via equation

2.8;

(2.8) 1 ℎ2 2 9.04 휆 = ( ) = 2푚퐸 1 퐸2 where 휆 is in Å and E is the energy of the neutron in meV and the mass m is taken as

1.675x10-27 kg. The wavelength is related to the velocity of neutron and thus the time of flight along a known path. Substituting in values in to equation 2.8 gives;

0.003955 0.003955 ∙ 푇 휆 = = (2.9) v 퐿

34 where v is in m(µs)-1 , T in µs and L in m. This gives a linear relationship between the neutron wavelength and the time of flight.

The High Resolution Powder Diffraction beamline (HRPD), as shown in Figure 2.5, was the instrument used to collect data in this work. HRPD is an example of a ToF beamline: The detectors are fixed at set angles and collect the scattered polychromatic neutrons over a period of time discriminating by time of arrival.

90° detectors (N side) Low angle detect ors (30°)

100 K CH4 moderator 100 m Ni guide

Disk choppe rs at 6 m & 9 m Beamstop

2 m sample position Backscattering detectors (168°) 1 m sample position

° 90° detectors (S side)

Figure 2. 5 - Schematic diagram of Isis HRPD 6

2.2-3 Rietveld refinement

The ‘Rietveld method’ is a process to obtain an accurate crystal structure from powder diffraction data using a “whole pattern fitting” method, each element of the powder pattern is broken down into a number of parameters, which can be refined or fixed by the user.

Firstly a crystallographic model is necessary which contains the lattice parameters, space group, atom positions and instrumental parameters such as the wavelength of radiation used.

This model is used to generate a calculated pattern, and the parameters are refined to minimise the difference between the observed and calculated pattern via a least-squares method74,77:

푛 2 (2.10) 푆푦 = ∑ 푤푖(푦푖 −푦푐푖) 푖=1

where 푆푦 is the residual, the quantity minimised by the refinement

푤푖 the weighting factor,

th 푦푖 and 푦푐푖 the observed and calculated intensities at the i step.

The calculated intensity is the sum of all Bragg reflections within a range:

2 푦푐푖 = 푠 ∑ 퐿퐾|퐹퐾 |∅푖 (2휃푖 − 2휃퐾)푃퐾퐴 − 푦푏푖 (2.11) 퐾 where;

s is the scale factor

A is the absorption factor

∅ is the reflection profile function

K represents the Miller indices hkl

LK is the Lorentz polarization factor

FK is the structure factor

PK is the preferred orientation function

ybi is the background intensity

The total profile function can be described as a convolution of instrumental and sample parameters:

 Axial and equatorial divergence;

 Emission profile;

 Sample size and shape broadening;

 Sample strain broadening.

The information contained within the peak shape (profile) is not always of consequence to obtaining a structural model and so three approaches to the modelling of the peak shape are used:

 Empirical peak shapes, where whichever function which best fits the data is used;

 Fundamental parameters, where the instrumental and sample contributions to the

profile are known and entered into the refinement;

 Semi-empirical, where the instrumental contributions are modelled empirically and

convoluted with sample contributions which can be refined;

In the case of empirical profile modelling, a convolution of Gaussian and Lorentzian functions with a number of refinable parameters, typically a pseudo-Voigt function, is used to describe the profile.

The mathematical agreement between the observed and calculated patterns can be ascertained by a number of functions:

1 2 ∑푖 푤푖[푦푖−푦푐푖 ] 2 (2.11) 푅푤푝 = { 2 } ∑푖 푤푖[푦푖]

1 (푛 − 푝) 2 (2.12) 푅 = { } 푒푥푝 ∑푛 푤 푦 2 푖−1 푖 푖

2 푅푤푝 휒2 = ( ) (2.13) 푅푒푥푝

where 푅푤푝 is the weighted profile function,

푅푒푥푝 is the expected R-factor

푤푖 the weighting factor

th 푦푖 and 푦푐푖 the observed and calculated intensities at the i step

However, these factors are only indicative of how good the fit is in a mathematical sense, it is possible for a local minima to have a good fit but make no chemical sense thus it is important to ensure that any structural model exhibits sensible bond distances and angles.

2.2-4 X-Ray Fluorescence Spectrophotometry

X-ray fluorescence is a technique to measure the elemental composition of a sample. Where

XRD involves the processes where X-rays are scattered by electrons, XRF detects processes which occur when X-rays excite electrons. When a core electron is excited by an X-ray a positive hole is left in the shell which can be filled by relaxation of an electron in a higher level shell due to attraction to the nucleus. This relaxation can result in the emission of a photon characteristic of that electronic transition78 as seen in Figure 2.6. The emission of these X-rays is governed by the following quantum mechanical selection rules79:

 Δn ≥ 1

 Δl = ±1

 Δj = ±1 where: n is the principal quantum number, l is the angular quantum number, j is the total momentum j=l+s; and s is the spin quantum number.

Based on these selection rules for the K series p→s transitions are allowed giving rise to two lines, the l series lines allow p→s, s→p and d→p transitions.

The photon emitted produces a spectral line which is named according to the original vacancy shell and how many shells above this the higher energy electron dropped down by; so an electron dropping to the K (1s) shell from the L(2s&2p) shell one level above would be termed the Kα line.

Figure 2. 6 - photoelectric effect resulting in fluorescence

The intensity of the line will correspond to the quantity of that element in the sample, thus it is necessary both to measure the intensity of lines as well as differentiate the different energies. There are two main approaches for this, energy dispersive or wavelength dispersive.

In the case of energy dispersive (EDX), a solid state Si detector is used which is capable of measuring both the intensity and energy level of emitted X-ray photons. The ionisation caused by the photons on the Si measured by a series of electrodes. The energy dispersive spectrometers typically have high count rates due to collecting a large portion of emitted photons, but poor energy resolution. EDX detectors can suffer from two particular artifacts, escape peaks and sum peaks. Escape peaks occur when X-rays undergo photoelectric

40 adsorption in-to the detector material and emit photoelectrons and X-rays characteristic of the detector material, commonly of 1.75 keV from Si detectors, which are picked up by the detector. Any fluorescence peaks of energy E are accompanied by escape peaks of energy E-

1.75 keV. Sum peaks occur when multiple X-ray photons collide with the detector simultaneously which creates an electron pule which is measured as the sum of these two collisions creating lines of multiples of the true fluorescence energy.

Wavelength dispersive (WDX) spectrometers use detectors which only measure the intensity of emissions and use a series of crystal monochromators and focusing optics in order to filter the beam and measure only one wavelength of X-rays at a time, this enables a greater energy resolution but due to the use of monochromators some intensity is lost.

Measured intensities depend not only on the concentration of the analyte element but also the concentration of all other elements present (matrix), the sample type (solid, liquid or powder), method of sample preparation, shape and thickness of the sample. A simplified equation for relating the measured intensity to the weight fraction of an analyte is as follows:

퐼푖 = 푘푖푊푖 + 푏푖 (2.14) where:

Ii is the radiation intensity

ki is a constant

Wi the weight fraction

bi is the radiation intensity when the analyte concentration is zero

41 ki and bi are fitted constants determined via a least squares method on the basis of reference samples.

Matrix effects such as absorption and enhancement are also important factors to consider and thus a new term is added, Mi, when greater than 1 this indicates absorption of radiation by the matrix to be the dominate effect. When smaller than 1 enhancement is the dominant effect.

Absorption effects occur when the fluorescence of the analyte element is sufficient to excite an element from the matrix, i.e. the absorption edge of a matrix element is slightly less than the characteristic line of an analyte element. The strength of this effect is determined by the relative absorption of the matrix and sample. If the matrix element has stronger absorption than the sample the effect is far greater. The enhancement effect or secondary fluorescence occurs when a matrix element emit radiation of slightly higher energy than the absorption edge of the analyte element.

In reality the equation for intensity is significantly more complex. If a sample is “infinitely thick” so that transmittance is zero the intensity equation79 can be written as:

휆푒푑푔푒 푑훺 휏푖(휆)퐼0(휆) 퐼푖 = 푄푖푞푖푊푖 ∫ (1 + ∑ 푊푗푆푖푗) 푑휆 (2.15) 4휋푠푖푛∅1 휒(휆′휆푖) 휆푚푖푛 푗

1 휏푖(휆푗) 휇(휆) 푠푖푛∅1 휇(휆푖) 푠푖푛∅2 푆푖푗 = 푞푗휏푗(휆) [푙푛 (1 + ) + 푙푛 (1 + ) ] (2.16) 2 휏푖(휆) 휇(휆푗)푠푖푛∅1 휇(휆) 휇(휆푗)푠푖푛∅2 휇(휆푖) where:

dΩ is the differential solid angle for the radiation, added as a geometric term for the irradiated surface area of the sample

qi is the sensitivity of the spectrometer for the characteristic line for analyte i

Wi and Wj are the weight fractions for analyte i and matrix element j

λmin and λedge are short wavelength limit and wavelength of the analyte absorption edge respectively, to integrate the area under the characteristic line

τ1(λ) is the photoelectric absorption coefficient for analyte i and wavelength λ, to take account of the absorption of the sample

Io(λ) is the intensity of the primary radiation

χ(λ, λi) , µ(λ) and µ(λi) are total mass-attenuation coefficients of the samples and specimens for incident and fluorescent radiation to account for the Cöester-Kronig transition probabilities

(where an electron from a higher subshell fills a vacancy in the same shell and an Auger electron is emitted) if the Lα or Lβ lines are used.

Sij is the matrix enhancement term as described above

Obviously the matrix effect is non-trivial but this shows how there is some dependence between the different compounds as the weight fractions rather than each analyte being treated individually with no consideration of other species present.

2.2-5 Scanning Electron Microscopy (SEM)

Similar to when an X-ray is incident upon a material, when an electron beam is used a number of effects are observed as depicted in Figure 2.7:

 Secondary electrons are emitted from which topological information can be gathered

 Auger electrons are emitted when a higher electron drops down to fill the hole left by

the excited electron, from which the surface sensitive composition can be examined as

these electrons are of characteristic energy of an element and would be reabsorbed by

the bulk sample;

 Backscattered electrons, which can give topological information;

 Characteristic X-rays are emitted from transitions in a similar fashion to XRF which

give compositional information;

 Cathodoluminescence – where an electron is incident upon a luminescent material and

results in the emission of a photon of energy in the visible region.

Figure 2. 7 - Penetration depth of electron beam and resultant emissions

Secondary electron detection is most commonly used for generating images. The number of secondary electrons detected is a function of the angle between the surface and the beam, thus images can be generated by the scanning across a sample with the electron beam in two dimensions. Emitted secondary electrons are often of lower energy (<50 eV) so that only those from the top surface (within a few nanometres) reach the detector before being reabsorbed.

Backscattered electrons (BSE) present another way to gather topographical information about a sample. These are higher energy electrons which undergo one or more scattering events and escapes from the surface with energy greater than 50 eV. In addition to topological information, BSE offers some degree of compositional information as well since the degree of backscattering is dependent upon which element the electron collides with, higher atomic numbers resulting in a greater proportion of backscattering due to a more highly charged nucleus. Whilst offering greater elemental contrast in images, BSE offer poorer resolution than secondary electrons. The higher energy of these electrons prevents them from being reabsorbed by the sample leading to a contribution from a larger sample area.

Characteristic X-rays are generated in a similar fashion to in XRF. A higher energy electron drops down in energy, emitting an X-ray photon in the process, to fill a positive hole left by a displaced electron excited by the incident beam, in this case an electron beam rather than X- ray beam. Energy rather than wavelength dispersive spectroscopy is more common when used in conjunction with SEM.

2.2-6 Ion Chromatography

Ion chromatography is a technique used to separate ions from a mixture in solution by using mobile and stationary phases which the ions interact with to different degrees.

The stationary phase consists of a solid medium presenting a charge opposite to that of the analyte ions so that the ions are attracted to the column. In the case of the system used in this work, a schematic shown in Figure 2.8, (Dionex ICS-1000) the column comprises a resin with quaternary ammonium ions.

Separation

column

Figure 2. 8 - Schematic of Ion chromatography system

The mobile phase is used to push the analyte through the system; in this work dilute methylsulphonic acid (~0.275%) was used as the eluent. The analyte ions will be bound to the stationary phase until the increase in concentration of the eluent becomes high enough to

46 exchange with them. This is referred to as the retention time of the ion; the more highly charged or polarizable a cation the longer the retention time.

Once through the column ions pass into a conductivity cell shown in Figure 2.9, where two stainless steel electrodes measure the conductance, in order to generate a signal.

The conductance of the eluent is minimised by the use of a suppressor, where water is electrolysed and forms O2 gas and hydronium cations in the anode chamber, and H2 gas and hydroxide ions in the cathode chamber. Due to an applied potential, eluent cations such as methylsulphonic acid (MSA) pass through a membrane towards the anode to combine with hydronium ions whilst the hydroxide ions in the cation chamber pass into the eluent chamber to neutralize the eluent reducing conductivity.

+ 2H3O +½O2

(g)

Figure 2. 9 - Schematic of conductance cell

2.2-7 Radiotracer studies

85Sr was used as a radiotracer in order to test ion exchange capacity within a highly dilute system. In order to examine the uptake, the activity of the ion exchange solution was measured before and after exchange. 85Sr decays primarily via electron-capture, emitting a gamma-ray of 514.01 keV with t½=64.84 days.

When gamma rays interact with matter three events may occur, the probability being dependent upon the energy of the incident gamma-ray and material:

 Photoelectric effect;

 Compton scattering;

 Pair production.

The photoelectric effect as described previously in the XRF section occurs most likely with lower energy gamma-rays, however, it should be noted that in addition to the emission of an

X-ray when an electron from a higher shell drops down, it is also possible for the emission of another low-energy electron, known as an Auger electron.

The Compton Effect occurs when the energy of the gamma-ray is high enough that it interacts with a valence orbital directly, rather than a core electron. In this case an electron is ejected from the atom to form an ion pair, and the gamma-ray is deflected at an angle with a lower energy:

퐸훾2 = 퐸훾1 − 퐼. 푃. −퐸푒

where

 퐸훾1 and 퐸훾2 are the energies of the incident and deflected gamma-rays respectively

 I.P. is the ionization potential of the ejected electron and

 Ee is the kinetic energy of the ejected electron

Pair production occurs when the gamma ray interacts with the coulomb field of the nucleus converting into an electron and a positron. The positron encounters another electron resulting in annihilation and two orthogonal gamma rays, each of 0.51 MeV. As according to 퐸 = 푚푐2, the minimum energy necessary to form an electron is 0.51 MeV, this process only occurs with gamma-rays above 1.02 MeV.

In this work, solid state semiconductor hyperpure Ge or NaI crystal scintillation detectors were used to measure the activity. These detectors use a scintillation material to produce photons via the photoelectric effect, Figure 2.10, which are then detected by a photomultiplier tube, Figure 2.11. When the photons interact with the semiconductor and form positive holes, the charges are separated due to an applied electric field and a pulse of collected charge can be recorded.

Figure 2. 10 - Example of photoelectric effect

Figure 2. 11 - Schematic of Ge semiconductor photomultiplier detector

Chapter 3: Titanosilicates and Zirconosilicates

3.1 Introduction

Many titanosilicates and zirconosilicates have been well studied and demonstrate good ion exchange capacity36,35,61. This chapter will focus on some the work done to attempt to synthesise new materials based on existing mineral phases for ion exchange applications.

Noonkanbahite, shown in Figure 3.1, is a natural mineral with the formula BaKNaTi2Si4O14, a member of the bakovite group of minerals which also includes batisite, BaNa2Ti2Si4O14, and

80 shcherbakovite, (K,Ba)KNaTi2Si4O14 . These minerals contain 4 SiO4 tetrahedral unbranched single-width chains within the unit cells which are connected by chains of TiO6 octahedra.

These materials appear to be interesting candidates for ion exchange due to the different cations in the structure. The divalent cations in the structure indicate a potential affinity for strontium cations; additionally different cation sites in the pores could potentially allow for uptake of both strontium and caesium without competition for the same site. The initial focus of this work was the synthesis of noonkanbahite or related natural minerals such as bakovite and shcherbakovite, which proved challenging. Whilst focus eventually shifted to other materials the synthesis of noonkanbahite remained a goal throughout the project.

Figure 3. 1 - Noonkanbahite along (100) showing connectivity of the polyhedra chains, with TiO octahedra in cyan, SiO4 tetrahedra in blue and sodium,, potassium and barium cations in yellow, purple and pink respectively

The synthesis of zirconium silicate materials has been well studied in particular the Na2O-

36 ZrO2-SiO2 system . However the addition of fluoride which acts as a mineralizer to increase the rate of dissolution of starting materials which can lead to the formation of different phases has not been reported so this was briefly investigated to see the effects.

Strontium zirconosilicates occur in nature, e.g. aqualite. This would indicate that hydrothermal synthesis of a strontium zirconosilicate should be possible. A strontium zirconosilicate could potentially be ion exchanged to remove the strontium and obtain the sodium or proton form, allowing for uptake of strontium into a lattice suited for strontium remediation. This chapter contains brief investigations into the synthesis of these zirconosilicate materials followed by the structural changes induced by initial ion exchanges and subsequent heat treatment of AV-3, a synthetic equivalent of petarasite.

3.2 Experimental

3.2-1 Noonkanbahite & related phases

Attempts to prepare a synthetic form of the natural mineral noonkanbahite type titanosilicate using a sol-gel type method were attempted after a significant number of attempts using a standard hydrothermal route yielded a mixture of amorphous glassy materials and simple silicate materials. Titanium isopropoxide(TIP) and Ludox® colloidal silica were mixed in

10ml ethanol solutions as in the table below, before being added to an aqueous solution of the group I and group II metal salts. The gels were aged for 40 minutes before heating hydrothermally at 200°C for 5 days.

Table 3. 1 – Titanosilicate synthesis gel preparations

TIP LUDOX® NaOH KOH Ba(OAc) Molar ratio Ref KF (g) 2 (g) HS-40 (g) (g) (g) (g) Ti:Si:Na:K:F:Ba

1 1.000 2.393 0.997 0 0.858 0 1: 4.5: 7.1 : 4.2: 4.2: 0

2 1.000 2.393 0 0.997 0 0 1: 4.5: 0 : 5.1: 0: 0

3 1.000 2.393 0.997 0 0 0.195 1: 4.5: 7.1 : 0: 0: 0.2

4 1.000 2.393 0.997 0 0.858 0.195 1: 4.5: 7.1 : 4.2: 4.2: 0.2

A switch was made from a sol gel type method to a simpler one with barium as the only extra- framework cation. The mixing of reagents was improved by using anhydrous titanium isopropoxide and tetraethylortho silicate (TEOS), in the amounts tabulated below, in 10ml ethanol to avoid premature hydrolysis. After thorough mixing of the solution, the barium

53 hydroxide solution was added to allow the hydrolysis before aging for 40 minutes and heating the gels hydrothermally at 200°C for 5 days.

Table 3. 2 - Titanosilicate synthesis gel attempts

Ref TIP (g) TEOS (g) Ba(OH)2·8H2O (g) H2O (g) Molar ratio Ti:Si:Ba:H2O

SG_1 1.000 1.066 0.700 18.504 1: 1.5: 0.6: 298

SG_2 1.000 1.218 0.800 21.147 1: 1.7: 0.7: 340

SG_3 1.000 1.421 0.933 24.671 1: 1.9: 0.8: 398

SG_4 1.000 1.705 1.120 29.606 1: 2.3: 1: 478

SG_5 1.000 2.132 1.399 37.008 1: 2.9: 1.3: 584

81 Returning to previously reported starting reagents of TiCl4 and colloidal silica, a number of gels were prepared by dissolving KOH, KF and NaOH in water according to the masses in

Table 3.3, followed by adding 1.0g TiCl4 in liquid form and Ludox® HS-40 collodial silica.

After homogenizing for 30 minutes the gel was heated the gels hydrothermally at 200°C for 5 days.

Attempts were also made to prepare synthetic shcherbakovite which is related to noonkanbahite but contains no barium. In this case the Si/Ti ratio was kept constant except in the case of sample H, in an effort to prepare the desired phase as shown in Table 3.3.

Table 3. 3 - Scherbakovite synthesis attempt gels

Ref TiCl4 (g) Ludox NaOH KOH (g) KF (g) Molar ratio (g) (g) Ti:Si:Na:K:F

A 1.000 1.335 5.003 3.472 0.654 1: 1.7: 23.7: 13.9: 2.1

B 1.000 1.335 2.497 1.733 0.654 1: 1.7: 11.8: 8: 2.1

C 1.000 1.335 5.003 2.315 0 1: 1.7: 23.7: 7.8: 0

D 1.000 1.335 2.497 1.733 0 1: 1.7: 11.8: 5.9: 0

E 1.000 1.335 5.003 3.472 0 1: 1.7: 23.7: 11.7: 0

F 1.000 1.335 3.361 2.315 0 1: 1.7: 15.9: 7.8: 0

G 1.000 1.322 0 6.945 0 1: 1.7: 15.4: 23.5: 0

H 1.000 2.670 5.003 3.472 0 1: 3.4: 23.7: 11.7: 0

Attempts to prepare a calcium titanosilicate involved dissolving Ca(OH)2 in H2O as in Table

3.4 followed by addition of titanium isoproxide and Ludox® HS-40 collodial silica allowing the mixture to gel. The gel was allowed to homogenize for 30 minutes before heating hydrothermally at 230°C for 5 days.

3.2-2 Titanite synthesis gels

Table 3. 4 - Titanite synthesis gels

Ref TIP (g) Ludox® Ca(OH)2 H2O (g) Molar Ratio (g) (g) Ti:Si:Ca:H2O

CaTixSiyOz A 0.570 2.567 1.000 28.558 0.15: 1.27: 1: 124

CaTixSiyOz B 1.061 2.572 1.000 19.039 0.28: 1.27: 1: 85

CaTixSiyOz C 0.820 0.856 1.000 19.039 0.21: 0.42: 1: 80

CaTixSiyOz D 0.950 1.711 1.000 19.039 0.25: 0.85: 1: 83

3.2-3 Zirconosilicate synthesis gels

In order to attempt the preparation of a strontium zirconosilicate, a series of gels was prepared according to Table 3.5, where ZrCl4 powder and Sr(OH)2·8H2O were dissolved in H2O before adding Ludox® HS-40 colloidal silica. Thick white gels were formed which were allowed to homogenize by stirring for 40 minutes before heating hydrothermally at 230°C for 5 days.

Table 3. 5 - Sr-Zr-Silicate synthesis attempts gel

Ref ZrCl4 LUDOX (g) Sr(OH)2·8H2O H2O (g) Molar Ratio (g) (g) Zr:Si:Sr:H2O

SrZrSiO2_1 1.000 1.000 0.548 37.020 1: 2.3: 0.5: 488

SrZrSiO2_2 1.000 1.000 1.245 37.020 1: 2.3: 1.1: 495

SrZrSiO2_3 1.000 1.000 2.191 37.020 1: 2.3: 1.9: 497

SrZrSiO2_4 1.000 1.000 2.448 37.020 1: 2.3: 2.1: 509

SrZrSiO2_5 1.000 1.000 4.191 37.020 1: 2.3: 3.7: 480

A final series of gels were prepared in an effort to investigate what effect, if any, fluoride would have on the synthesis of sodium zirconosilicates. Gels were prepared according to

3 Table 3.6 where ZrCl4 and NaF powders along with NaOH pellets were dissolved in 20 cm

H2O before adding Ludox® HS-40 collodial silica. Thick white gels were formed which were allowed to homogenize by stirring for 40 minutes before heating hydrothermally at 230°C for

5 days.

Table 3. 6 - Na-Zr-SiO2 -F synthesis gel preperations

Ref ZrCl4 (g) LUDOX (g) NaOH (g) NaF (g) Molar Ratio Zr:SiO2:Na:F

Zr_Si_NaF A 1.000 1.00 2.50240 1.111 1: 1.55: 14.7: 6.2

Zr_Si_NaF B 1.000 3.00 2.50240 1.111 1: 4.66: 14.7: 6.2

Zr_Si_NaF C 1.000 9.00 2.50240 1.111 1: 13.98: 14.7: 6.2

AV-3 was prepared as per literature82 by dissolving 1.43g NaOH pellets, 2g NaCl, 1g KCl and

3 0.84g ZrCl4 powder in 7.21 cm deionized water. To this was added 5.35 g sodium silicate solution (23% m/m SiO2, 8% m/m Na2O). The gel was then allowed to homogenize before heating under autogenous pressure for 10 days at 230°C.

3.2-4 Ion exchanges

3 1.00g of ground AV-3 was added to 25cm of Sr(NO3)2 solutions of concentrations

0.001 mol-1 dm3, 0.05 mol-1 dm3, 0.5 mol-1 dm3 or 1.0 mol-1 dm3, before shaking for 24 hours. The resultant suspension was filtered, washed well with deionized water and the solids formed into fused beads for XRF analysis.

3.3 Results and discussion

3.2-1 Noonkanbahite and related phases

Preparing precursor gels as an aqueous mixture using colloidal silica and water-soluble metal salts did not produce a material with a Ti-O-Si framework , only TiO2 and SiO2 phases as seen in Figure 3.2. This may be due to the rapid gelling of the silica as it hydrolyses rapidly in a basic environment. A very thick gel was produced; where the TIP did not appear to mix properly forming an inhomogeneous emulsion rather than the homogeneous colloidal sol as in a typical sol-gel process.

4500 * 1Molar ratio 2 3 4000 Ti:Si:Na:K:F:Ba 4

* * 3500 * * * 4 * 1: 4.5: 7.1 : 4.2: 4.2: 0.2 3000

3 2500

* * * * 1: 4.5: 7.1 : 0: 0: 0.2

Counts * 2000 ￭￭ × 2 1500 1: 4.5: 0: 5.1: 0: 0

1000 1

500 ￭￭ 1: 4.5: 7.1 : 4.2: 4.2: 0 ×

0 10 20 30 40 50 60

2 Theta (°)

Figure 3. 2 – XRD patterns of the first synthesis attempt of a new titanosilicate * - TiO2 Anatase – PD F 01-086-1157

￭ - TiO2 – PDF 04-001-1352

× - SiO2 - PDF 01-089-1667

The sol-gel type method did not yield any interesting phases either, as seen in Figure 3.3, the

phases are predominately amorphous material with a small amount of crystalline TiO2. There

is a small trace of a crystalline phase, however due to very low intensity peaks and poor signal

to noise ratio this assignment is somewhat dubious.

4500 1 2Molar ratio 3 4000 5Ti:Si:Ba:H2O * 6 3500 ￭ SG_5

3000 1: 2.9: 1.3: 584

2500 SG_4 1: 2.3: 1: 478 Counts 2000 SG_3

1500 1: 1.9: 0.8: 398

1000 SG_2

1: 1.7: 0.7: 340 500 × SG_1 0 10 20 30 40 50 60 1: 1.5: 0.6: 298 2 Theta (°)

Figure 3. 3 - XRD patterns of sol-gel attempts at noonkanbahite

* - Ba2Si3O8 – PDF 04-011-1317

￭ - TiO2 – PDF 01-070-2556

× - BaTiSiO5 – PDF 00-011-0150

12000 AMolar ratio B C * DTi:Si:Na:K:F E F G 10000 H

8000 * * * * * * * * * H 1: 3.4: 23.7: 11.7: 0 ￭￮￭ 6000 G 1: 1.7: 15.4: 23.5: 0

Counts × ￮ × × ￮￮￮ × × ×× × F 1: 1.7: 15.9: 7.8: 0 +

4000 E 1: 1.7: 23.7: 11.7: 0

D 1: 1.7: 11.8: 5.9: 0

2000 C 1: 1.7: 23.7: 7.8: 0

B 1: 1.7: 11.8: 8: 2.1

A 1: 1.7: 23.7: 13.9: 2.1 0 10 15 20 25 30 35 40 45 50 2 Theta (°)

Figure 3. 4 - XRD patterns of scherbakovite attempts Molar ratio

* - Na2Ti(SiO4)O Natisite PDF – 04-011-0023 Ti:Si:Na:K:F ￭ - KCl PDF – 04-002-3657

×- K2TiSi3O9(H2O) Umbite PDF – 04-012-4486

￮ - Na2TiSi5O13 ETS-10 PDF -04-013-2831 + - SiO2 - PDF 01-089-1667

H 1:1.7:15.9:11.7:0 Figure 3.4 shows natisite was the major crystalline phase observed in this series identified in

samples A,C & H. However natisite has been shown to be a poor ion exchanger83 unless

modified due to sodium cations being trapped between layers of TiO. However, more recently

some work has been undertaken by Readman et al. where doping Zr in to the natisite structure

was shown to improve Cs uptake slightly 84. Sample F shows of umbite along with ETS-10,

both known and well-studied titanosilicate materials.

3.2-2 Titanite

Molar ratio 4000 A Ti:Si:Ca:H O B 2 * * C 3500 * D * ￭￭￭ * * CaTixSiyOz_D ￭￭￭￭ * * * 3000 0.25: 0.85: 1: 83

2500

CaTixSiyOz_C

2000 0.21: 0.42: 1: 80

Counts

1500 CaTixSiyOz_B

1000 0.28: 1.27: 1: 85

500 CaTixSiyOz_A

0 0.15: 1.27: 1: 124 10 20 30 40 50 60 2 Theta(°)

Figure 3. 5 - XRD patterns of CaOH - TiO2-SiO2 system syntheses

* - CaTi(SiO4)O Titanite – PDF 04-013-2626 ￭ - Ca (Ti2O4(OH)2) Kassite – PDF 01-072-6913

Molar ratio

Ti:Si:Ca:H2O 62

CaTixSiyOz_D In addition to a lack of work reported on the subject on calcium titanosilicates, 90Sr is known

to replace calcium in bone tissue85 so in order to obtain a material which could potentially

selectively uptake strontium , a material with exchangeable extra-framework calcium cations

was a synthesis target. Initial synthesis attempts, shown in Figure 3.5, through hydrothermal,

sol-gel and solid state methods yielded mostly amorphous end phases. Only hydrothermal

synthesis with titanium as the framework metal resulted in any crystalline products, of these

mixed phase systems the titanite phase was observed via a pattern search and match against

the PDF database. Titanite was later prepared again as a pure poorly crystalline phase via a

hydrothermal synthesis route. Rietveld refinement of this phase, from data acquired on a

Bruker D8 in reflection geometry using a structure from the PDF database86 as a starting

model, Figures 3.6 & Table 3.7, yielded a structure shown in Figure 3.7 with selected bond

distances and angles shown in Table 3.8.

200000 Observed CalculatedCalculatCaclulateded Difference

150000

100000

Counts 50000

-50000 10 20 30 40 50 60 70 2 Theta (°) Figure 3. 6 - Rietveld refinement of Ca-titanite using TOPAS Academic 4.3

Figure 3. 7 - Refined structure of Ca-titanite with blue SiO4 tetrahedra, cyan TiO6 octahedra and Ca spheres.

Table 3. 7 - Rietveld refinement parameters of Ca-titanite occupancies and displacement parameters (Beq) were both fixed at 1 Å².

a /Å 7.0337(4) X-ray source Cu Kα1 and Kα2

b /Å 8.7125(6) Scan time 8 hours

c /Å 6.5534(5) Step size 0.02°

V /Å3 367.78(7) Background Chebyshev

β /° 113.74(7) Peak shape profile Tompson-Cox Hastings

space group P21/a

Z 4

Nº independent parameters 50

Rwp 4.781

Rexp 0.487

Rp 2.863

Site x y z

Ca1 0.217(1) 0.4175(5) 0.251(2)

Ti1 0.501(2) 0.2499(9) 0.743(2)

Si1 0.749(3) 0.4354(8) 0.253(4)

O1 0.716(4) 0.305(1) 0.753(4)

O2 0.920(4) 0.326(3) 0.450(5)

O3 0.087(4) 0.183(3) 0.074(5)

O4 0.396(4) 0.472(3) 0.668(4)

O5 0.593(3) 0.031(3) 0.851(5)

Table 3. 8 - Bond distances and angles from Ca-titanite Rietveld refinement

Distance Bond Bond Angle ° Bond Angle ° (Å)

Ca1-O1 2.46(2) O1-Ti-O1 173.3(2) O1-Ca-O2 137.4(2)

Ca1-O2 3.05(1) O1-Ti-O2 89.6(1) O1-Ca-O3 152.1(1)

Ca1-O2 2.40(1) O1-Ti-O3 84.8(2) O1-Ca-O4 58.2(1)

Ca1-O3 2.50(1) O1-Ti-O4 85.6(2) O1-Ca-O4' 78.1(2)

Ca1-O4 2.54(1) O1-Ti-O5 88.6(1) O1-Ca-O5 80.6(1)

Ca1-O4 2.75(1) O2-Ti-O3 172.6(3) O1-Ca-O5' 74.7(2)

Ca1-O5 2.41(2) O2-Ti-O4 94.96(9) O2-Ca-O3 64.9(2)

Ca1-O5 2.33(2) O2-Ti-O5 94.7(1) O2-Ca-O4 84.0(1)

Ca1-Si1 3.06(2) O2-Ca-O4' 74.2(1)

O2-Si-O3 105.5(2) O2-Ca-O5 122.7(1)

Ti1-O1 2.166(9) O2-Si-O4 116.6(2) O2-Ca-O5' 143.2(2)

Ti1-O2 1.93(1) O2-Si-O5 114.5(1) O3-Ca-O4 124.5(1)

Ti1-O3 2.06(1) O3-Si-O4 91.6(2) O3-Ca-O4' 129.7(1)

Ti1-O4 2.12(1) O3-Si-O5 110.5(1) O3-Ca-O5 71.7(2)

Ti1-O5 1.996(9) O4-Si-O5 115.0(1) O3-Ca-O5' 93.0(1)

O4-Ca-O4' 76.4(1)

Si1-O2 1.56(1) O4-Ca-O5 91.6(2)

Si1-O3 1.74(1) O4-Ca-O5' 132.0(1)

Si1-O4 1.55(2) O5-Ca-O5' 71.2(2)

Si1-O5 1.75(1)

Si1-Ca1 3.06(1)

The refined cell and average Ca-O, Ti-O and Si-O bond distances all agree with those reported in the literature86,87 which are predominantly data derived from single crystal diffraction experiments on natural titanite samples. The agreement of atom positions between literature and the refined data is acceptable, although there is some displacement of the cations, considering the poor quality of the sample as indicated by the wide peaks shown in

Figure 3.6, which prevented a good fit from being obtained.

Titanite

1200 100.02

0.00 1000 100.00 -0.02 800 -0.04 99.98 600 -0.06

exo

Temp/C 400 Mass/% 99.96 -0.08

200 -0.10 99.94 0 -0.12

-0.14 99.92 0 20 40 60 80 100 120 Time/min

Time/min vs Temp/C Time/min vs DTA/(mW/mg) Time/min vs Mass/%

Figure 3. 8 TGA trace of Ca-titanite between 30 and 1000 °C under N2

Titanite shows good thermal stability; powder XRD of a sample heated in a furnace to

1000°C indicates no notable change to the structure and TGA shows little change in mass loss or the DTA trace. A drop of 0.04% at approx. 700°C is the most significant event which could

67 be due to the loss of some small amount of volatile impurity present. A small change is evident between 200 and 300°C which coincides with a structural change from P21/a to A2/a as described by Taylor and Brown87.

Unfortunately titanite did not show any significant strontium ion exchange behaviour from initial ion exchange experiments analysed by XRF and preliminary 85Sr ion exchange work, as described in Chapter 5 section 3.4.

3.3-3 Fresnoite

Attempts to produce a new barium titanite phase yielded primarily amorphous products and mixtures of simple condensed materials; the only product of note was a synthetic analogue to the natural mineral fresnoite (Ba2TiSi2O8) with a small amount of an unknown impurity.

Fresnoite is a porous layered titanosilicate containing titanium in an unusual 5 coordinate square pyramidal geometry. Unfortunately, as in the case of titanite, no ion exchange behaviour was observed during preliminary ion exchange experiments studied using XRF. A

Rietveld refinement, Figures 3.9 & 3.10, Table 3.9 & 3.10, was obtained from data acquired using a Bruker D8 diffractometer in reflection mode using a structure from the PDF database

88 as a starting point . Fresnoite shares some structural characteristics with natisite (Na2TiSiO5) such as the square pyramidal TiO5 units and both materials being layered structures. In this case it may not be surprising that fresnoite does not exchange well as the cations are locked in place by the layered structure.

6000 Observed CaclulatedCalculated Difference 5000

4000

3000

Counts 2000

1000

-1000 10 20 30 40 50 60 2 Theta (°)

Figure 3. 9 - Rietveld refinement of fresnoite phase with excluded areas

Table 3. 9 - Rietveld refinement parameters and atom positions for Fresnoite refinement where occupancies and displacement parameters (Beq) were fixed at 1 Å²

a /Å 8.5188(2) X-rays Cu Kα1 and Kα2

c /Å 5.2067(2) Scan time 3 hours

V /Å3 377.85(3) Step size 0.02°

space group P4bm Background Chebyshev

Peak shape profile Thompson-Cox

Z 8 Hastings

Nº independent parameters 21

Rwp 9.777

Rexp 5.086

Rp 6.877

excluded region 1 30.5-32.2

excluded region 2 23.6-24.4

excluded region 3 38.25-39.24

Site x y z

Ti1 0 0 0.444(7)

Ba1 0.329(9) 0.828(8) 0

Si1 0.127(3) 0.626(8) 0.483(1)

O1 0.371(6) 0.549(5) 0.578(2)

O2 0.135(5) 0.635(2) 0.288(8)

O3 0 0.5 0.659(2)

O4 0 0 0.445(2)

Figure 3. 10 - Refined structure of fresnoite viewed along (001) above and (010) below with TiO5 in cyan, SiO4 in blue and barium cations in brown

Table 3. 10 - Selected refined bond distances and angles of fresnoite

Angle 1 Angle 2 Angle 3

Bond Distance (Å) Bond (deg) (deg) (deg)

Ti1 - O4 1.70(9) O1-Ti-O4 106.2(2)

Ti1 - O1 1.98(6) O1-Ti1-O1 85.6(7) 148.0(8)

Ba1 - O2 2.77(4) O1-Si1-O1 104.3(5)

Ba1 - O1 2.82(5) O1-Si1-O2 117.0(4)

Ba1 - O3 2.83(4) O1-Si1-O3 103.8(4)

Si1 - O2 1.59(7) O1-Ba1-O1 55.0(4) 96.6(5)

Si1 - O3 1.62(7) O1-Ba1-O2 71.9(5) 118.3(3) 163.8(4)

Si1 - O1 1.62(6) O1-Ba1-O3 52.2(3) 91.0(5)

Si1 - Si1 3.02(4) O2-Ba1-O3 74.1(5) 160.8(6)

Despite the sample being impure, the fit can be compared acceptably in terms of unit cell and average bond distances to the literature88 where single crystal data was collected, however, the

R-factors in the overall fit are high as are the errors in the atomic positions. Whilst the atomic positions are reasonably close, with the exception of the titanium atom, which is displaced by approximately 0.5 Å along the z direction, the model generated is only reliable in terms of backing up the assignment of the main phase produced as fresnoite rather than as a definitive crystal structure.

3.3-4 Zirconosilicate synthesis

Whilst much work has been done on the synthetic phases of the NaOH–ZrO2–SiO2 system, the introduction of fluoride to act as a mineralizing agent may alter which phases crystallize

and if a novel phase was possible. Furthermore, the use of ZrCl4 instead of ZrO2 may also

result in a difference in the observed phases for a given composition. Figure 3.11 shows the

formation of various Zr:Si ratios with the addition of fluoride. In the case of all of these

syntheses NaF is present in large amounts; the formation of a simple salt such as this would

likely occur rapidly. The phases produced, Na2ZrSi2O7·H2O and Na2ZrSi3O9·3H2O were not

novel and have been studied for ion exchange previously in literature45,47 so no further steps to

obtain a pure phase were taken.

6000 1:1Molar Ratio 1:3 1:9Zr:SiO2:Na:F × 5000 × × Zr_Si_NaF C × 1:1.6:14.7:6.2 4000 ￭￭￭￭￮￭ * ￭￭ * ￭ * 3000 * Zr_Si_NaF B Counts ￭ * ￭￭￭￭￭￭ 1:4.7:15:6.2 2000

Zr_Si_NaF A 1000 1:14:15.2:6.2

0 10 20 30 40 50 60 2 Theta (°)

Figure 3. 11 - XRD patterns of the ZrO2-SiO2-NaOH-NaF system × - NaF – PDF 00-004-0793 ￮ - SiO2 – PDF 01-075-3900 Zr_Si_NaF A ￭ - Na2ZrSi2O7·H2O – PDF 04-017-2269 * - Na2ZrSi3O9·3H2O (Gaidonnayite) PDF 04-012-8668 1:14:15.2:6.2

Molar Ratio

Zr:Si:Sr:H2O 12000 1 2 3 SrZrSiO2_5 * 4 5 10000 1:1.6:3.7:516.4 × * * * * * × × ×× * 8000 SrZrSiO2_4 ￭ 1:1.6:2.2:504.2

6000 Counts SrZrSiO2_3 1:1.6:1.9:502.4 4000 ￭ SrZrSiO2_2 ￭￭￭￭￭￭￭￭￭￭￭￭￭￭￭￭ 1:1.6:1.1:495.8 2000 * SrZrSiO2_1 0 ￭ 10 20 30 40 50 60 1:1.6:0.5:490.9 2 Theta(°)

Figure 3. 12 - XRD patterns of synthesis attempts at Sr-zirconosilicate

× - SrCl2·6H2O PDF 04-010-2886

Molar Ratio ￭ - SrCl2·2H2O PDF 00-025-0891

* - ZrSiO4 (Zircon) PDF 01-076-0865 Zr:Si:Sr:H2O

SrZrSiO2_5

An attempt was made to prepare a strontium zirconosilicate, a strontium-containing metal 1:1.6:3.7:516.4 silicate that could potentially then be acid-exchanged under highly forcing conditions to SrZrSiO2_4 remove the strontium leaving a potentially strontium specific material. However, under these 1:1.6:2.2:504.2 conditions the formation of zircon was too favourable to produce any other phases with the

SrZrSiO2_3 strontium forming hydrated chloride salts, including SrCl2·6H2O which should be too soluble

to collect. The observation of the hexahydrate salt may be due to the hydrothermal1:1.6:1.9:502.4 bombs

leaking water vapour which resulted in a supersaturated gel, allowing the salt to crystallize. SrZrSiO2_2 The formation of only zircon and not a strontium zircon silicate could possibly be due to the 1:1.6:1.1:495.8

Sr(OH)2·8H2O not being as strong a base or as soluble as the typical NaOH/KOH for these SrZrSiO2_1 74 1:1.6:0.5:490.9

types of syntheses. Adding another base such as tetrabutylammonium hydroxide was thought to be able to lower the pH of the starting gel to disfavour the formation of zircon however hydrothermal heating of these gels resulted in amorphous materials. Whilst it may have been possible to investigate further the possibility of creating this strontium phase, a thorough investigation was deemed to be too time and equipment intensive and a broad approach was taken to synthesis attempts.

3.3-4a- Synthesis and characterisation of synthetic petarasite (AV-3)

AV-3 is a synthetic analogue to the natural zirconosilicate petarasite with the formula

Na5Zr2Si6O18(OH, Cl) ·2H2O. AV-3 was reproduced using hydrothermal methods according to work by Lin et al.66 and the structure refined against data acquired from a Bruker D8 X-ray diffractometer in reflection geometry ( Figure 3.13& 3.14 and Tables 3.11-3.13). The initial model used for the refinement was from the PDF database82.

12000 Observed CalculatedCaclulated [TypDifference 10000 e a 8000 quote 6000 from 4000

Intensity arb.u the 2000 docu 0 ment -2000 10 20 30 40 50 60 2 Theta or the sum Figure 3. 13 - Rietveld refinement of as made petarasite (AV-3) using TOPAS academic V4.1

mary

of an

intere

sting

point.

You

can

positi

the

text

box

anyw

here Figure 3. 14 - Refined structure of a petarasite (AV-3) along (001) Blue tetrahedra show SiO4 units and green octahedra ZrO6 units. Yellow spheres indicate Na+ cations, green Cl- anions and grey water molecules in the

docu 76 ment.

Use

2 Table 3. 11 - Rietveld refinement parameters of petarasite (AV-3) displacement parameters (Beq) were fixed at 1 Å

X-ray Cu Kα1 and Kα2 a / Å 10.785(9) Scan time 8 hours b / Å 14.42(2) Step size 0.02° c / Å 6.708(5) Background Chebyshev 3 V / Å 944.3(2) Peak shape Thompson-Cox

profile Hastings β / ° 113.24(6) space group P21/m

Z 4 NO. independent parameters 107

Rwp 5.609

Rexp 2.441

Rp 3.975 Site x y z occ Na1 0.506(3) 0.087(2) 0.883(5) 0.96(4) Na2 0.944(2) 0.130(2) 0.018(4) 0.88(4) Na3 0.794(3) 0.25 0.483(7) 1.06(4) Zr 0.7532(7) 0.009(5) 0.511(2) 1 Si1 0.032(2) 0.141(2) 0.577(4) 1 Si2 0.290(2) 0.046(2) 0.993(4) 1 Si3 0.448(2) 0.151(2) 0.368(3) 1 O1 0.062(5) 0.25 0.562(1) 1 O2 0.920(3) 0.114(2) 0.586(8) 1 O3 0.102(4) 0.068(3) 0.467(8) 1 O4 0.134(4) 0.100(2) 0.881(9) 1 O5 0.762(3) 0.027(3) 0.882(7) 1 O6 0.695(3) 0.038(3) 0.233(6) 1 O7 0.755(4) 0.120(3) 0.122(9) 1 O8 0.640(3) 0.123(3) 0.430(6) 1 O9 0.438(4) 0.25 0.482(9) 1 O10 0.422(3) 0.084(2) 0.517(9) 1 Wat1 0.143(4) 0.25 0.257(2) 0.8(4) Wat2 0.489(5) 0.25 0.883(1) 0.95(6) Cl 0.801(2) 0.25 0.036(6) 0.78(4) O-H1 1.051(4) 0.25 0.187(2) 0.9(4) O-H2 0.7393(4) 0.25 0.094(1) 0.70(8)

Table 3. 12 Selected bond distances from refinement of AV3

Bond Distance (Å) Bond Valence Bond Distance (Å) Na1_O5 2.97(7) 0.042 Zr_O2 2.13(7) Na1_O6 2.49(5) 0.156 Zr_O2 2.37(4) Na1_O6 2.96(6) 0.044 Zr_O5 2.45(5) Na1_O7 2.80(4) 0.068 Zr_O5 2.51(4) Na1_O10 2.21(3) 0.334 Zr_O6 1.76(7) Na1_Wat2 2.29(5) 0.267 Zr_O6 1.85(5)

Na1_Cl 3.72(7) Zr_O8 2.17(8)

Na1_O-H2 3.08(3) 0.032 Zr_O8 2.43(8) SUM 0.942 Si1_O1 1.66 (7) Na2_O2 2.52 (7) 0.143 Si1_O2 1.36(8) Na2_O3 3.13(3) 0.028 Si1_O3 1.51(5)

Na2_O4 2.38(7) 0.211 Si1_O4 1.93(3)

Na2_O5 2.40(8) 0.201 Na2_O5 3.02(6) 0.037 Si2_O4 1.74 (7) Na2_Wat1 2.72(6) 0.083 Si2_O7 1.20(7) Na2_Cl 2.30(5) Na2_O-H1 2.15(7) 0.391 Si3_O7 1.66 (5) SUM 1.093 Si3_O8 1.71(5) Si3_O9 1.70(3) Na3_O1 2.68 (5) 0.094 Si3_O10 1.56(7) Na3_O2 2.26(6) 0.289 Na3_O8 2.47(3) 0.167

Na3_Cl 3.02(7) 0.038

Na3_Cl 3.60(4) Na3_O-H2 2.36(7) 0 .222 SUM 0.810

Table 3. 13 Selected bond angles from refinement of AV3

Bond Angle(°) Bond Angle(°) O2_Zr_O2 94.6(8) O1_Si1_O2 115.3(7) O2_Zr_O5 87.5(5) O1_Si1_O3 112.4(8) O2_Zr_O6 91.9(7) O1_Si1_O4 105.8(8) O2_Zr_O6 113.5(6) O2_Si1_O3 124.8(4) O2_Zr_O8 83.6(6) O2_Si1_O4 92.7(6)

O2_Zr_O8 167.1(8) O3_Si1_O4 99.5(6)

O2_Zr_O5 95.8(8) O2_Zr_O6 106.5(4) Si2_Si2_O4 113.6(7) O2_Zr_O6 82.4(6) Si2_Si2_O7 135.7(8) O2_Zr_O8 174.9(5) O4_Si2_O7 98.9(6) O2_Zr_O8 73.2(7) O5_Zr_O6 157.7(8) O5_Zr_O6 158.9(6) O7_Si3_O8 115.5(4) O5_Zr_O8 88.9(7) O7_Si3_O9 121.4(8) O5_Zr_O8 97.8(6) O7_Si3_O10 101.9(5) O6_Zr_O6 31.0(6) O8_Si3_O9 111.4(6) O6_Zr_O8 68.9(7) O8_Si3_O10 106.6(8) O6_Zr_O8 87.6(8) O9_Si3_O10 96.3(4) O6_Zr_O8 93.9(4)

Whilst the overall structure matches well with that reported in literature 82, some bond distances shown in Table 3.12 appear to be erroneous. The Zr-O6 bond distances reported are too small at 1.79 Å and 1.85 Å as the expected average bond distance is ~2 Å; This may be due to some disorder in the structure or, more likely given the reasonably high errors in this refinement, that the O6 position is inaccurate. The bond valence sums of the cation positions are approximately 1, which is what is expected for a monovalent cation.

After obtaining a pure product, AV-3 was added to solutions of varying concentrations of

Sr(NO3)2 in order to examine any ion exchange capacity as shown in Table 3.14.

Table 3. 14 - Compositions of solids, reported as oxides as measured by XRF after initial exchange experiments

→Sr(NO3)2 Parent 0.01 M 0.05 M 0.5 M 1.0 M concentration ↓Compound

ZrO2 49.29 57.18 60.61 46.95 54.12

SiO2 31.58 29.62 25.79 28.78 28.68

SrO 0.00 5.49 4.85 13.34 10.77

Na2O 15.26 4.14 4.67 7.91 3.56

Cl 3.87 3.57 4.07 3.01 2.89

Elemental analysis of the solid after exchange by XRF indicated the presence of strontium in the material with a decrease in the levels of sodium, which is a good indication of ion exchange occurring and not adsorption on the surface of the material of strontium salts.

SEM and EDX, Figures 3.15 & 3.16, were used to examine the crystal size and shape and to check for the presence of any strontium-containing amorphous material which would not be evident from X-ray diffraction but would likely appear as globular or irregularly shaped materials on SEM images. The SEM used to acquire these images was a Philips XL-30 FEG

Environmental SEM with EDS. To acquire images an accelerating voltage of 15 keV with a spot size of 3, for EDX the spot size was increased to 5 and a voltage of 20 keV was used.

Figure 3. 15 - SEM image of AV-3 as a powder dispersed over carbon tape followed by platinum sputter coating

Figure 3. 16 - SEM image of AV-3 as a powder dispersed over carbon tape followed by platinum sputter coating

The parent material can be seen to be composed of plate-like crystals clustered together, with no obvious signs of amorphous material. EDX over an area indicates the average composition

After ion exchange the plate-like crystals are maintained with no morphology changes or amorphous material evident. As seen in Figure 3.17 the strontium peak in the EDX cannot be distinguished from the silicon peak due to the overlapping emission energies.

Figure 3. 17 - EDX spectrum of SEM images indicating Sr/Si overlap

Initially a powder X-ray diffraction pattern for the strontium-exchanged material was obtained with a lab-based Bruker D8 diffractometer which resulted in poor quality data with very broad peaks similar to the previous pattern recorded as shown in Figure 3.13, however, later high quality synchrotron data was obtained from the I11 beamline at the Diamond Light Source which allowed for a much better fit to be obtained. The data in Figure 3.18 were acquired using the I11 MAC detectors with a wavelength of 0.825704(2) Å and a zero-point of

0.000792(2)°.

16000 Observed CalculatedCaclulated 14000 Difference

12000

10000

8000

Counts 6000

4000

2000

-2000 10 20 30 40 50 60 70 80 2 Theta (°)

Figure 3. 18 - Rietveld refinement of AV-3 performed using TOPAS 4.1 Academic

Figure 3. 19 - Refined structure of Sr-exchanged AV-3 viewed along (001) with blue SiO tetrahedral, green ZrO6 octahedra, yellow Na+ spheres, large green Cl- spheres, small green Sr2+ spheres and red and grey water spheres.

Table 3. 15 - Rietveld refinement parameters for AV-3 displacement parameters were fixed

a /Å 10.801(1) b /Å 14.494(2) Background Chebyshev c /Å 6.613(7) β /° 113.251(8) Peak shape profile Thompson-Cox V /Å3 951.9(8) Hastings

space group P21/m Z 4 NO. independent parameters 115

Rwp 4.17

Rexp 0.562

Rp 3.073 Site x y z occ B eq Na1 0.526(6) 0.136(6) 0.989(7) 0.620(4) 1.0 Na2 0.509(4) 0.006(6) -0.089(9) 0.444(5) 1.0 Na3 0.964(9) 0.25 0.568(9) 0.436(7) 1.0 Sr1 0.925(9) 0.116(2) 0.917(4) 0.401(1) 1.0 Sr2 0.806(9) 0.215(9) 0.024(9) 0.101(1) 1.0 Sr3 0.759(3) 0.25 0.608(5) 0.436(1) 1.0 Zr 0.749(1) 0.015(1) 0.514(2) 1 1.0(1) Si1 0.043(2) 0.126(2) 0.593(7) 1 0.7(6) Si2 0.254(3) 0.055(2) 1.007(9) 1 1.2(5) Si3 0.459(2) 0.157(2) 0.4(5) 1 0.7(4) O1 0.06(4) 0.25 0.622(9) 1 1.2(3) O2 0.902(3) 0.122(4) 0.436(7) 1 1.8(5) O3 0.096(4) 0.108(4) 0.454(9) 1 1.1(9) O4 0.145(3) 0.121(3) 0.876(7) 1 0.7(8) O5 0.787(4) 0.02(4) 0.862(9) 1 0.5(7) O6 0.698(3) 0.043(4) 0.049(9) 1 1.0(4) O7 0.393(3) 0.05(3) 0.289(7) 1 0.7(8) O8 0.64(2) 0.111(5) 0.431(8) 1 0.8(5) O9 0.441(3) 0.25 0.386(9) 1 1.1(5) O10 0.409(4) 0.097(4) 0.584(7) 1 1.0(6) Wat1 0.078(3) 0.25 0.131(9) 1 1.1(7) Wat2 0.375(5) 0.25 0.878(8) 1 1.2(6) Cl 0.732(5) 0.25 0.063(6) 0.651(7) 1.0 O-H1 0.778(9) 0.25 -0.163(7) 0.338(9 1.0 O-H2 0.821(9) 0.25 0.432(9) 0.562(7) 1.0

Table 3. 16 - Refined bond distances and valence sums for AV-3

Bond Valence Bond Distance (Å) Bond Distance (Å) Bond Valence Sums Sums

Na1-O6 2.22(7) 0.33 Sr1-Wat2 2.34(6)

Na1-O7 2.33(6) 0.24 Sr1-O10 2.35(5) 0.538

Na1-O10 2.42(6) 0.19 Sr1-O6 2.45(5) 0.41

Na1-O6 2.54(8) 0.13 Sr1-O7 2.57(5) 0.294

Na1-Wat2 2.81(8) Sr1-O6 3.11(8) 0.069

Na1-O7 3.10(8) 0.03 Sr1-O-H1 3.34(9)

Na1-O8 3.42(7) 0.01 Sr1-O9 3.41(7) 0.03

Na1-O5 3.59(6) 0.01 Sr1-O8 3.48(7) 0.025

Total 0.94 Sr1-O8 3.54(6) 0.022

Sr1-O7 3.56(6) 0.02

Sr1-O5 3.60(5) 0.018

Sr1-O4 3.92(7) 0.008

Total 1.434

Bond Valence Bond Distance (Å) Bond Distance (Å) Bond Valence Sums Sums Na2-O5 2.00(6) 0.58 Sr2-Wat1 2.12(6)

Na2-O3 2.50(6) 0.15 Sr2-O-H1 2.45(5)

Na2-Wat1 2.78(8) Sr2-O5 2.48(5) 0.38

Na2-O-H1 2.79(8) Sr2-O4 2.65(7) 0.24

Na2-O6 2.86(7) 0.06 Sr2-O3 2.73(6) 0.19

Na2-O2 3.12(7) 0.03 Sr2-O2 2.75(6) 0.18

Na2-O5 3.15(5) 0.03 Sr2-O5 3.19(5) 0.06

Na2-O4 3.17(8) 0.02 Sr2-O-H2 3.46(8)

Na2-O4 3.32(6) 0.02 Sr2-O6 3.53(9) 0.02

Na2-O-H2 3.45(8) Sr2-O1 3.80(8) 0.01

Total 0.89 Sr2-O8 3.86(5) 0.01

Sr2-O1 3.95(4) 0.01

Sr2-O2 3.96(6) 0.01

Total 1.11

Bond Distance (Å) Bond Valence Bond Distance (Å) Bond Valence

Na3-O8 2.45(7) 0.18 Sr3-O8 2.04(7) 1.24

Na3-O8 2.45(8) 0.18 Sr3-O8 2.04(7) 1.24

Na3-O-H2 2.70(6) 0.09 Sr3-O-H1 2.20(7)

Na3-O2 2.79(6) 0.07 Sr3-O2 2.53(8) 0.33

Na3-O2 2.79(5) 0.07 Sr3-O2 2.53(5) 0.33

Na3-Wat1 3.23(8) 0.02 Sr3-O9 3.22(9) 0.05

Na3-O9 3.45(7) 0.01 Sr3-O1 3.24(4) 0.05

Na3-O1 3.54(6) 0.01 Total 3.24

Total 0.62

Table 3. 17- Refined bond angles for AV-3

Bond Angle ° Bond Angle ° O2-Zr-O3 85.3(3) O1-Si1-O2 121.0(3) O2-Zr-O5 86.2(3) O1-Si1-O3 114.6(3) O2-Zr-O6 92.7(4) O1-Si1-O4 102.3(4) O2-Zr-O8 82.4(2) O2-Si1-O3 110.9(4) O2-Zr-O10 171.6(2) O2-Si1-O4 100.2(5) O3-Zr-O5 84.6(1) O3-Si1-O4 104.7(3) O3-Zr-O6 94.44(9) O3-Zr-O8 167.6(2) O4-Si2-O5 108.4(4) O3-Zr-O10 95.4(1) O4-Si2-O6 120.0(2) O5-Zr-O6 178.6(2) O4-Si2-O7 105.2(1) O5-Zr-O8 93.2(2) O5-Si2-O6 107.4(1) O5-Zr-O10 85.5(3) O5-Si2-O7 108.1(2) O6-Zr-O8 87.5(3) O6-Si2-O7 107.3(3) O6-Zr-10 95.6(2) O8-Zr-O10 96.3(4) O7-Si3-O8 107.4(1) O7-Si3-O9 112.2(2) O7-Si3-O10 112.5(2) O8-Si3-O9 98.8(4) O8-Si3-O10 114.9(4) O9-Si3-O10 110.3(2)

When examining Tables 3.16 and 3.17 it should be considered that X-ray diffraction and the

Rietveld method have been used to derive the structure thus the sites are averaged and so whereas there appears to be two cations in each pore it is more likely the case there is only one cation with equal occupancies over the two equivalent positions. In the case of the cations within the pore they refined as if all the species present in the pore void of the framework—Na, Sr, Cl, OH and H2O— were in the pore at the same time but have fractional occupancies. When introducing Sr2+ into the structure a sodium atom will need to be removed, this may cause the framework around this vacant site to shift to move closer to another cation within the pore. The averaging of defects in the framework caused by missing cations may be the cause of the apparent shorter than expected bond

86 distances observed, such as Na2-O5, however, it is difficult to confirm this and further work would be useful.

For this refinement Na and Sr sites were initially constrained to be on the same positions with occupancies fixed to ensure charge balancing; as the refinement became more stable these constraints were relaxed. This method, whilst helping to ensure a stable refinement, can limit the movement of cations within the pore and in addition it assumes all sites will exchange equally, which is not likely to be the case.

The Sr3 sites in particular have bond valence sums which appear normally high, due to duplicate

Sr3-O8 bonds which are unusually small, implying that the Sr2+ cations would not reside in the Sr3 site due to the proximity to O8. However, the refined occupancies indicate that the Sr3 site does contain a high portion of the exchanged strontium. In this case if the refined occupancies are to be believed it is likely that either cation site position is not exactly correct and the true bond distances are larger, or that the framework differs when the pore contains strontium compared to when the pore contains sodium and this discrepancy causes an apparently small bond which does not exist in reality.

AV-3

1000 0.1 102

0.0 101 800

-0.1 100 600

-0.2 99

400

exo

Temp/C -0.3 98 Mass/%

200 -0.4 97

0 -0.5 96

-0.6 95 0 20 40 60 80 100 120 Time/min

Time/min vs Temp/C Time/min vs DTA/(uV/mg) Time/min vs Mass/%

Sr AV-3

1000 0.1 102

0.0 800 100

-0.1

600 98 -0.2

400 -0.3 96

exo

Temp/C

Mass/% -0.4 200 94 -0.5

0 92 -0.6

-0.7 90 0 20 40 60 80 100 120 Time/min

Time/min vs Temp/C Time/min vs DTA/(uV/mg) Time/min vs Mass/%

Figure 3. 20 - TGA traces of AV3 (above) and Sr-AV3 (below) including DTA trace

By examining the TGA data shown in Figure 3.20, strontium exchange appears to lead to a

change in the strength of structurally bound water based, on the DTA trace and temperature of

water loss observed. This is also observed in Chapter 4 when examining TGA data of Sr-AV7.

The mass loss due to water occurring from 200°C is more rapid in the strontium-exchanged

sample compared to the pristine material and the DTA trace indicates a more intense

endothermic event in the pristine material.

30C 900C 35000 1000C 1100C Cool back to 30C 30000 Cooled to 30°C 25000

20000 1100°C

Counts

15000 1000°C

10000 900°C

5000 30°C

10 15 20 25 30 35 40 45 50 2 Theta

Figure 3. 21 - Variable temperature XRD patterns of AV-3 heated from room temperature to 1000 °C followed by cooling back to room temperature

High temperature treatment could potentially lead to collapsed structures with more desirable

low leaching properties; to study this samples have been heat treated slowly at 1 °C/min using an

Anton-Parr heating stage coupled to a Bruker D8 X-ray diffractometer. Sr-AV3 undergoes a

phase transition to a trinclinic phase (P-1) parakeldyshite when heated to high temperatures as

shown in Figure 3.21. The phase change is non-reversible. It is unclear whether this proceeds via

an intermediate phase or gradual transformation with an expansion in the unit cell as expected

from heating. No impurity Sr-containing phases have been detected after cooling, suggesting any

strontium remains in the structure or is in an amorphous phase.

The parakeldyshite phase was refined as shown in Figure 3.22 to obtain a model shown in Figure

3.23; data was acquired from the I11 beamline at Diamond light source using the MAC detector

and a wavelength of 0.825704(2) Å and a zero-point of 0.000792(2)°.

35000 Observed CalculatedCaclulated 30000 Difference[Ty

25000 pe a

20000 quot 15000

Counts 10000 e

5000 fro

0 m -5000 the -10000 10 20 30 40 50 60 2 Theta (°) doc

ume Figure 3. 22 - Rietveld refinement of parakeldyshite performed using TOPAS academic 4.1 from synchrotron data acquired from the I-11 beamline at the Diamond light source nt or

the

In comparison to AV-3, parakeldyshite has lost some of the SiO4 connectivity. Ins umAV-3 the

tetrahedra form rings of 6 units whereas in parakeldyshite the tetrahedra form pairsmar connected by

ZrO6 octahedra. In comparing the bond valence sums shown in Tables 3.16 & 3.y19 of, site Sr1 and

Sr2 in AV-3 both have less strong binding between the Sr2+ cation and the frameworkan in

comparison to parakeldyshite, 1.434 & 1.11 vs 2.46 &1.88, although the Sr3 siteinter in AV-3, which

does not exist in parakeldyshite, remains the strongest binding site for Sr2+ with aesti bond valence

sum of 3.24, which is significantly higher than the expected value of 2. The higherng than expected

90 poin

t. value is due to the close proximity of 4 structural oxygen atoms in the framework, although it is possible that the Sr3 position is in fact an averaged position and the Sr2+ cation in this site has equal occupancy in positions closer to one pair of these four oxygen atoms. Refining the displacement parameter may have been able to reveal this, however, due to the correlation to occupancy it was decided to fix this parameter in favour of refining the occupancy. It should also be noted that the Zr-O6 bond distance appears abnormally long, ~2.2 Å compared to an average of ~2 Å. Whether this is an artefact of the refinement or a real bond distance is unclear; as this is a high temperature phase it would not be unreasonable to expect some distortions or disorder.

Figure 3. 23 - Refined structure of Sr-parakeldyshite - high temperature phase from AV-3 viewed along the c axis with ZrO6 in green, SiO4 in blue and sodium and strontium cations in yellow and green, respectively.

Table 3. 18 - Rietveld refinement details of parakeldyshite

a /Å 6.674(1) α /° 92.35(2) b /Å 8.829(2) β /° 93.78(1) c / Å 5.433(1) γ /° 71.09(2) V /Å3 302.1(1) space group P-1 Background Chebyshev Z 2 Peak shape Thompson-Cox NO. independent parameters 100 profile Hastings

Rwp 6.264

Rexp 1.078

Rp 4.79

Site x y z occ Beq Zr1 0.288(2) 0.2673(2) 0.2237(3) 1 1.08(8) Si1 0.6442(7) 0.1542(5) 0.769(8) 1 1.27(8) Si2 0.9329(7) 0.3305(5) 0.6865(9) 1 1.30(7) Na1 0.859(3) 0.078(2) 0.251(2) 0.55(3) 1 Na2 0.34(5) 0.512(4) 0.817(9) 0.39(5) 1 Sr1 0.912(3) 0.121(3) 0.28(3) 0.12(8) 1 Sr2 0.356(2) 0.4965(2) 0.756(5) 0.22(4) 1 O1 0.289(1) 0.0392(8) 0.175(1) 1 1.11(2) O2 0.876(1) 0.1794(8) 0.738(1) 1 0.75(2) O3 0.4988(9) 0.2219(7) 0.521(1) 1 1.26(9) O4 0.552(1) 0.2444(9) 0.017(1) 1 1.1(1) O5 0.016(1) 0.3121(8) 0.417(1) 1 0.7(2) O6 0.111(1) 0.3376(8) 0.853(1) 1 1.0(2) O7 0.243(1) 0.5207(7) 0.2687(1) 1 0.6(2)

Table 3. 19 - Selected bond distances from the refined structure of parakeldyshite

Bond Bond Bond Distance(Å) Bond Distance(Å) Bond Distance(Å) Valence Valence Zr1 -O1 2.02(7) Na1-O1 2.75(2) 0.08 Sr1-O1 2.48(2) 0.38 Zr1 -O3 2.03(6) Na1-O1' 2.84(2) 0.06 Sr1-O2 2.53(2) 0.33 Zr1 -O4 2.09(7) Na1-O2 2.38(2) 0.21 Sr1-O2' 2.57(2) 0.29 Zr1 -O5 2.08(8) Na1-O2' 2.77(2) 0.07 Sr1-O2'' 2.99(2) 0.09 Zr1 -O6 2.27(6) Na1-O2'' 2.98(2) 0.04 Sr1-O3 2.98(2) 0.10 Zr1 -O7 2.17(8) Na1-O3 2.804(2) 0.07 Sr1-O4 2.64(2) 0.24 Na1-O4 2.41(2) 0.19 Sr1-O5 2.11(3) 1.02 Si1-O1 1.65(8) Na1-O5 2.71(2) 0.09 Total 2.46 Si1-O2 1.55(9) Total 0.81 Sr2 -O3 2.62(2) 0.26 Si1-O3 1.63(7) Na2 -O3 2.90(5) 0.05 Sr2-O4 2.61(2) 0.26 Si1-O4 1.59(9) Na2-O4 2.59(4) 0.12 Sr2-O4' 2.61(2) 0.26 Na2-O5 2.66(4) 0.10 Sr2-O5 2.65(2) 0.24 Si2 -O2 1.65(9) Na2-O6 2.52(4) 0.14 Sr2-O6 2.57(2) 0.29 Si2-O5 1.59(9) Na2-O7 2.57(6) 0.13 Sr2-O7 2.64(2) 0.24 Si2-O6 1.46(9) Na2-O7' 2.77(4) 0.07 Sr2-O7' 2.70(3) 0.21 Si2-O7 1.48(7) Na2-O7'' 3.00(6) 0.04 Sr2-O7'' 2.92(3) 0.11 0.65 Total 1.88

Table 3. 20 - Selected framework bond angles from the refined structure of parakeldyshite

Bond Angle ° Bond Angle ° Bond Angle ° O1-Zr-O3 94.69(3) O1-Si1-O2 102.77(4) O2-Si2-O5 107.96(4) O1-Zr-O4 95.26(3) O1-Si1-O3 107.55(5) O2-Si2-O6 109.48(3) O3-Zr-O6 166.61(3) O1-Si1-O4 118.46(6) O2-Si2-O7 112.41(4) O4-Zr-O5 174.56(6) O2-Si1-O3 106.71(7) O5-Si2-O6 105.88(3) O1-Zr-O7 172.46(4) O2-Si1-O4 107.55(4) O5-Si2-O7 114.40(3) O4-Zr-O7 171.47(6) O3-Si1-O4 114.5(3) O6-Si2-O7 106.42(6) O4-Zr-O8 88.86(4)

3.4 Conclusions

In this chapter brief investigations aimed at preparing new titano and zirconosilicate materials were made, with a hydrothermal synthesis and characterization of titanite and attempts at preparing ion-exchanged forms reported. Titanite appears not to undergo ion exchange easily due to strong binding of the calcium cation although computational modelling would help support this. Synthetic petarasite-AV-3 has been prepared along with an ion-exchanged counterpart and the structural changes through ion exchange and heat treatment investigated. Upon heat treatment AV-3 can be seen to transform in to a denser parakeldyshite phase.

Chapter 4: Tin silicates

4.1 Introduction

As previously discussed in chapter 1 little work has been undertaken on the synthesis of tin silicate materials and less on the ion exchange properties of these materials. The majority of papers are based on the structure of mineralogical samples although some work on synthesis was done on behalf of the Exxon Research and Engineering group89; e.g. Lin et al. described the structure of a synthetic tin variant of the natural mineral kostylevite, AV-7, as described in

Chapter 1. This work was followed up with a report on the structure of another tin silicate90 Sn

AV-11, K4SnSi6O18 , a rhombohedral system with similar extended chains of –O-Si-O- units as in AV-7 and umbite. AV-11can be produced directly or by the high temperature treatment of

AV-7. High temperature treatment of materials commonly leads to more dense phases and so the heat treatment of ion-exchanged materials has the potential to lead to more ideal waste forms for long term storage with lower porosity and ion exchange properties. Tin has two common oxidation states, II and IV, the latter being the more stable. Adjusting the oxidation state of tin could potentially lead to new materials and/or change the ion exchange properties due to a change in the charge of the framework and thus the attraction between framework and cations.

This chapter contains results from exploratory synthesis in the sodium-tin-silicate system.

Conditions similar to those used to produce AV-7 by Lin et al. 37 were used as a starting point.

Preliminary ion exchange work on AV-7 followed along with the structural changes caused by this ion exchange and subsequent heat treatments.

4.2 Experimental

4.2-1 Hydrothermal SnO2-SiO2-NaOH synthesis survey

Informed by the synthesis conditions of AV-7, a brief survey in varying compositions was attempted to find possible new materials. Ten different ratios (Table 4.1) of SiO2 (colloidal silica,

3 Ludox® HS-40), SnO2 and NaOH were combined by dissolving NaOH and SnO2 in 20 cm water before adding Ludox® HS-40 to form gels before homogenizing for 40 minutes then heating hydrothermally under autogenous pressure in Parr acid digestion vessels to 230°C for one week.

Table 4. 1 - Mass compositions (g) of SiO2-SnO2-NaOH survey

Ref SnO2 SiO2 NaOH Mole Ratio Sn_Si_1 1 5.02 0.60 1: 2: 0.9 Sn_Si_2 1 5.02 1.26 1: 2: 1.9 Sn_Si_3 1 5.02 2.83 1: 2: 4.2 Sn_Si_4 1 5.02 4.85 1: 2: 7.3 Sn_Si_5 1 5.02 7.54 1: 2: 11.3 Sn_Si_6 1 5.02 11.30 1: 2: 17 Sn_Si_7 1 6.27 1.88 1: 2.5: 2.8 Sn_Si_8 1 6.27 3.23 1: 2: 4.8 Sn_Si_9 1 6.27 5.02 1: 2.5: 7.5 Sn_Si_10 1 6.27 7.54 1: 2.5: 11.3 Sn_Si_11 1 4.18 1.57 1: 1.67: 2.4 Sn_Si_12 1 4.18 2.69 1: 1.67: 4.0 Sn_Si_13 1 4.18 4.19 1: 1.67: 6.3 Sn_Si_14 1 4.18 6.28 1: 1.67: 9.4

4.2-2 Microwave TinII survey

A brief investigation into forming mixed sodium SnII/SnIV silicate species, altering the ratio of

SnII : SnIV whilst maintaining the Na:Si:Sn ratio of 30:5.7:1 was made. A high proportion of

NaOH was used to promote hydrolysis of the tin and silicate species, along with high amounts of silica to prevent formation of tin species with no sodium or silicon.

Reagents were combined into a gel according to the compositions in table 4.2 below, by

3 ® dissolving NaOH, SnCl2 and SnCl4·5H2O in 20 cm water before adding Luodx HS-40 and colloidal silica and homogenizing for 40 minutes before heating to 210°C for 10 h under autogenous pressure using a CEM MARS6 microwave system.

Table 4. 2 – Mass Compositions (g) for microwave SnII Survey

colloidal Ref SnCl ·5H O SnCl NaOH Mole ratio 4 2 2 silica II Sn _1 1.75 0.3 2.7 4 1: 0.3: 9.0: 20.0 II Sn _2 2.63 0.15 2.7 4 1: 0.1: 6.0: 13.3 II Sn _3 0.8 0.45 2.7 4 1: 1.0: 19.7: 43.8 II Sn _4 0 0.6 2.7 4 0: 1: 14.2: 31.6

II Following this SnCl2 was replaced with Sn oxalate and the compositions in table 3 below were made up into precursor gels by dissolving NaOH and SnII oxalate in 30 cm3 water followed by adding Ludox® HS-40 colloidal silica. The precursor gels were allowed to homogenize for 30 minutes before heating under autogenous pressure to 210°C for 10 h using a CEM MARS6 microwave system.

Table 4. 3 - Compositions (g) for soidum SnII oxalate silicates synthesis gels

SnII colloidal Mole Ref NaOH Oxalate silica ratio A 2.067 4.5 0.4 01:03:01 B 2.067 4.5 0.8 01:03:02 C 2.067 4.5 1.2 01:03:03

4.2-3 AV-7 - synthetic tin kostylvite

AV-7 was prepared by dissolving 2.16g KF powder and 4.08g NaOH pellets in 10 cm3 deionized water. To this was added 3.25 g SnCl4·5H2O which was stirred until fully dissolved followed by

10.19 g Ludox® HS-40 colloidal silica as opposed to the metasilicate used in the Lin et al.37 method. The mixture was then allowed to homogenize for 40 minutes before heating under autogenous pressure for 5 days at 230°C.

4.2-4 Ion exchanges

4.2-4a Bulk ion exchange

3 -1 1.00 g of ground AV-7 was added to 25cm of Sr(NO3)2 solutions of concentrations 0.001 mol dm3, 0.05 mol-1 dm3, 0.5 mol-1 dm3 and 1.0 mol-1 dm3, before shaking for 24 hours. The resultant suspensions were filtered, washed well with deionized water and the solids formed into fused beads for XRF analysis.

4.2-4b Low concentration ion exchange

3 0.02 g of AV-7 was added to 10 cm of 50 ppm of Sr(NO3)2 solution in triplicate before shaking for 24 hours before filtering to examine the filtrate by ion chromatography. Each ion chromatography sample is run in triplicate and the session is calibrated with 8 standards of purchased K+, Na+ and Sr2+ 1000 ppm solutions made up to 5 ppm, 10 ppm, 20 ppm, 30 ppm, 40 ppm, 60 ppm, 100 ppm and 200 ppm of each cation.

4.2-4c Fused bead XRF

Fused beads of each sample were prepared by grinding 0.15 g sample with 3.0 g of mixed

64%/36% lithium metaborate / tetraborate flux. The mixture was then added to a gold-platinum alloy crucible and heated to 1050°C for 7 minutes before removing and adding a small amount

(<0.1 g) of ammonium iodide to act as a releasing agent to prevent the glass bead from sticking to the platinum. After re-heating for a further 7 minutes the crucible was removed again then swirled to remove air bubbles. The glass was heated for a further 2 minutes to re-melt the glass and form a smooth even surface before finally cooling and removing the bead from the crucible.

This particular flux and high dilution were used due to the propensity of tin to form inclusions within the fused bead matrix. XRF data was collected on an Bruker S8 Tiger spectrometer using a calibrated method using standards prepared with fused beads. The measurement method used a customized method for this system where each element was measured for either 10 or 100 seconds depending on the intensity of the Kα1 emission lines used, the total measurement time being approximately 8 minutes a sample.

4.3 Results and discussion

4.3-1 SnO2-SiO2-NaOH synthesis survey

Whilst the literature contains a great deal of work done on the mapping of the SiO2 - ZrO2 -

59 NaOH system , little work has been done on the SnO2 – SiO2 – NaOH system. In this series of experiments the hydrothermal method was used with constant time and temperature throughout the series. It is evident that under all conditions, when starting from SnO2, no metal silicate formation is taking place as cassiterite, SnO2 adopting the rutile structure, is observed throughout all experiments with various silica and sodium silicate phases as seen in Figure 4.1. It was first considered that the SnO2 was not decomposing due to lack of a mineralizing agent such as KF, however, these experiments were run with the addition of KF by a year 4 project student, Ryan

George and a similar trend was observed. In comparison to the procedure to prepare AV-7

(NaKSnSi3O9·H2O) where SnCl4·5H2O is used as the source of tin which dissolves readily in water to form hydrated salts, SnO2 is insoluble in dilute base, requiring a fused alkali hydroxide to form an intermediate, and will not hydrolyse without harsher reaction conditions such as even lower pH or higher temperatures. The rate of hydrolysis of the metal salt vs. the rate of hydrolysis of SiO2 and indeed the exact identity of the hydrated metal species in the reaction gel is likely an important issue in the formation of these types of metal silicates as the reaction between the M-OH and Si-OH species.

100

5500

5000 x

4500 4000 x 3500 3000 x

Counts 2500 2000 * 1500 x 1000 x ** *+ 500 + + + x 0 10 20 30 40 50 60 2 Theta (°)

Figure 4. 1 - Typical PXRD pattern from NaOH-SnO2-SiO2 series

* - SiO2 PDF 04-012-2128

+ - γ-Na2Si2O5 PDF 00-018-1242

x- SnO2 PDF 04-003-0649

4.3-2 Microwave TinII survey

TinII metal silicates are uncommon due to the favourable oxidation of SnII to SnIV, under

standard hydrothermal conditions, resulting in formation of SnO2 over several days. However,

due to the rapid uniform heating conditions of microwave synthesis SnII can be used to form a

stannosilicate before oxidation occurs, whilst the exact reason for the prevention of oxidation

from heating via microwaves is not yet understood it has been proposed that due to SnIIO

IV 91 absorbing microwave radiation much more intensely than Sn O2 , which may also be the case

for the hydrated tin hydroxide species thought to be present in the synthesis gels. Thus

microwave heating of precursor gels containing SnII were attempted in an effort to produce new

phases.

101

Initially a mixture of SnII and SnIV chlorides were in varying ratios used in the precursor gels along with NaOH and colloidal silica before microwave treatment. Despite varying the ratio of

SnII:SnIV there was no great difference in the observed phases, as seen in Figure 4.2, tin- containing phases appearing to be various chlorides and silicon-containing phases to be primarily sodium silicate type phases however, there are a number of unidentified peaks which may correspond to a stannosilicate material although indexing the pattern was not successful.

6000 ₒ + + 5000 ₒ ￭ + + 4000 ￭￭ + ￭ + ₒ ₒ + ￭ * 3000

Counts

2000

1000

0 10 15 20 25 30 35 40 45 50 2Theta (°)

Figure 4. 2 - Diffraction patterns of SnII:SnIV chloride microwave synthesis attempts

￭ - δ-Na2Si2O5 PDF 00-053-1234

ₒ - γ-Na2Si2O5 PDF 04-014-8492

+ - Na2(SiO2(OH)2) ·8H2O PDF 00-060-0360

Upon switching to a different source of SnII, tin oxalate, a new phase was prepared across a variety of Sn:Si ratios and pH conditions which may indicate the new phase is a thermodynamically stable, easy to form phase. No structure matching this diffraction pattern has been found in the PDF database, nor any structure which matches the cell which has been indexed as face centred cubic system, with a=10.39 Å. Using space group of Fm-3m, one of five

102

in the F--- extinction group, a Pawley fit, Figure 4.3, has been performed which indicates good

agreement between the predicted peak positions and observed without an excessive number of

unobserved reflections. The structure of the new material has yet to be solved, however ion

exchange experiments have been performed and no observed uptake has been shown.

45000 Observed

Caclulated 40000 Calculated Difference

35000

30000

25000

20000

Counts 15000

10000

5000

-5000

-10000

10 20 30 40 50 60 70 80 90

2 theta(°)

Figure 4. 3 - Pawley refinement of new Sn-Si-O phase using Fm-3m space group

HF digestion and subsequent ICP analysis performed by Warwick analytical service gave a

composition of 35.79% Sn and 25.62% Si giving an estimated atomic ratio of 1:3 Sn:Si. Solid

state NMR data was also obtained from the EPSRC National Solid-state NMR Service at

Durham for both 29Si and 119Sn. 119Sn NMR proved difficult. The interpretation provided

indicated a very weak signal was located around -200 & -300ppm with a large chemical shift

anisotropy suggesting a single SnII environment rather than a SnIV species, as this would show a

103 chemical shift of ~-600ppm with less anisotropy. It should be noted, however, that the data quality is very poor and this interpretation is somewhat tenuous.

The 29Si NMR signal, Figure 4.4, was collected using a recycle time of 60 seconds, spin-rate of

6000 Hz and a pulse duration of 6.4 µs. Whilst the signal is still reasonably broad indicating the material is not highly ordered, it can be resolved into 3 component peaks at -111.5 ppm, -101.7 ppm and -92.7 ppm, corresponding to Q4, Q3 and Q2 tetrahedral silicon, respectively. The most intense peak is the Q4 peak at -111.5ppm which indicates the majority of silicon in the framework is bound through oxygen atoms to four other silicon atoms. The abundance of Si-O-

Si bonds would suggest either Sn atoms are clustered in a siliceous framework or the majority of the Sn atoms are in a different phase to the silica framework.

104

Chemical shift

Figure 4. 4 - 29Si NMR of unknown Sn-Si-O sample

Attempts to solve the structure have proceeded by three main routes:

Attempting to build manually an initial model by adding atoms to the Wyckoff positions of one of five possible space groups whilst maintaining the correct Si:Sn ratio and using a rough assumed volume per atom of ~18Å3, which is a commonly used guideline for non-hydrogen atoms92.

Generation of an initial model through charge flipping followed by simulated annealing using the

Rietveld method.

105

Using the EXPO software which attempts to solve the structure using the biased random model generation method in direct space.

After the NMR data revealed the structure may have disorder, low temperature X-ray diffraction experiments were performed to try to reduce the thermal motion of the structure and thus reduce the degree of disorder.

An initial laboratory diffractometer scan at 100K revealed asymmetrical splitting of certain peaks: 002, 022, 004, 402, 422, 044, 622

Indexing the new pattern using EXPO, which uses TREOR90 to index, yielded the following cell with space group I2 a non-standard description of space group No. 5: a & c = 7.353 Å b =

10.300 Å β = 91.466°.

However upon acquiring higher quality synchrotron data at 100 K from the I11 beamline at the

Diamond Light Source, indexing yielded a new cell with primitive centring which allows for a more sensible transformation of the cell. A Pawley fit, Figure 4.5, was performed using this new data fitting to the new cell; P42/n with a = 7.2981(9) Å and c = 10.5675(4) Å.

106

20000 Observed CalculatedCaclulated Difference 15000

10000

5000

Counts

-5000

-10000 10 20 30 40 50 60 70 80 90 2 theta (°)

Figure 4. 5 - Pawley fit of Sn-Si-O sample at 100 K using synchrotron data

Thermal gravimetric analysis, Figure 4.6, of the sample under O2/N2 indicates mass loss over a

wide temperature range but using the coupled mass spectrometer this can be broken down in to

multiple steps. Loss of water is seen in two steps, from 30-80°C followed by 150-200°C. Water

loss at lower temperature would likely be due to surface water on the sample, but higher

temperature water loss would indicate the presence of structural water in the system. Loss of

CO2 indicates the presence of an organic compound as heating in O2 should lead to complete

combustion. As the synthesis of this material involves tin oxalate it is possible that the oxalate

species is involved in the structure. XRF of the materials using a semi-quantitative method

calibrated against standards, running each sample for 18 minutes, gave a composition of 42.44%

O, 30.87% Sn, 29.27% Si and 1.83% Na. Based upon the elemental analysis by ICP-MS and

XRF and TGA a possible composition would be NaSn3Si9O25·0.5C2O4·H2O which maintains

107 charge balance from oxalate with a small amount of sodium, which corresponds to approximately 1% by mass.

1.4e-10 1.5 102

1.2e-10 1.0 100

1.0e-10 0.5 98 8.0e-11 0.0 6.0e-11 96

Exo

-0.5 Mass % Ion Current Ion 4.0e-11 94 -1.0 2.0e-11 92 -1.5 0.0

-2.0 90 0 200 400 600 800 Temp (°C)

Temp./oC vs DTA/(mW/mg) Type equation here. Temp./oC vs Mass/% Temp./oC vs M18 Ion Current Temp./oC vs M 44 Ion Current

Figure 4. 6 - TGA plot with mass spectrometry signals of H2O (blue) and CO2 (green)

A range of information has been gathered about this material including some structural information from XRD and NMR together with compositional information from TGA/MS and

ICP/MS, all of this information together should lead to the eventual solving of this structure but work is still on-going.

108

4.3-3 AV-7 - synthetic tin kostylvite

AV-7 is a stannosilicate material based upon the natural mineral Kostylvite first reported in a patent application by Corcoran89 et al. for the Exxon group, but the structure was not solved until

2000 by Lin et al.60 Due to the open hydrated framework containing both Na+ and K+ and structural similarity to K-umbite, a material shown to perform well as an ion-exchanger49, this material was thought of as a potential ion exchange material itself.

AV-7 was synthesised according the literature with a change in the source of silica, with colloidal silica (Ludox HS-40) being used in place of sodium metasilicate due to availability and ease of preparation as it removes the need to dissolve the metasilicate during the preparation of the precursor gel. This change did not seem to affect the morphology of the sample when comparing SEM images.

4.3-4 Ion Exchanges

Following the successfully reproduced synthesis of AV-7, initial ion exchange experiments were

3 performed. 1.00g of ground AV-7 was added to 25cm of Sr(NO3)2 solutions of varying concentrations and shaken for 24 hours, the resultant suspensions were filtered, washed well with deionized water and the solid formed into fused beads for XRF analysis as shown in Table 4.4.

Unfortunately XRF analysis for this material is not as accurate as possible due to problems with the sample preparation method. In forming a fused bead the material is heated to over 1000°C in

109 the presence of a lithium borate flux in order to form a more diffuse and heterogeneous sample with a smooth surface. In this case the tin in the sample is oxidised to form SnO2 which then forms an inclusion in the glass bead reducing the effectiveness of this technique. More routine sample preparations such as pressed pellets also proved to be inaccurate due to the heavy Sn atoms absorbing fluorescence from neighbouring lighter atoms.

Table 4. 4 - Compositions by percentage mass of solid samples derived from XRF

Material Na K Sr Sn Si

Parent NaKSnSi3O9 7.3(1) 23.0(1) 0 24.00(6) 15.7(3) 0.01M 5.1(1) 22.9(1) 2.77(4) 24.72(6) 16.5(3) 0.05M 4.4(1) 13.9(1) 14.28(4) 19.68(6) 14.2(3) 0.5M 5.4(1) 13.4(1) 15.19(4) 18.95(6) 13.8(3) 1.0M 5.4(1) 12.6(1) 15.34(4) 17.42(6) 13.3(3)

Despite the lack of completely accurate results obtained from XRF, it is still evident from the data that some strontium is either being incorporated in to the material through ion exchange, adsorbing on the surface of the material or being precipitated out as an insoluble phase.

Through XRD no new phases were observed after ion exchange which would be due to a crystalline insoluble material and no higher background or areas of obvious amorphous material were observed, which would indicate the strontium has ion exchanged into the material.

SEM/EDX was performed to look for possible amorphous impurities; as seen below AV-7 can be seen form small cuboidal crystallites which tend to aggregate together in a flower-like cluster.

110

Figure 4. 7 - EDX (below) from above area of SEM (above) image with automatic peak assignment

111

SEM & EDX in Figure 4.7 indicates the presence of gold due to the conductive coating applied to the sample. The automatic assignment software also indicates the a large amount Ti due to the overlap of O. This amount of Ti is unlikely as an impurity would not have shown in up in the multiple scans of different regions of the sample. Unfortunately many of the elements present have emission lines which overlap, in this case it can be seen the potassium emission lines (Ka1

3.31 keV, Ka2 3.31 keV, Kb1 3.59 keV) overlap with Sn (La1 3.44 keV, La2 3.43 keV, Lb1 3.66 keV)

After ion exchange the sample appears to be mostly unchanged in terms of the morphology, with

AV-7 remaining as aggregates of small cuboidal particles. EDX results cannot indicate the presence of strontium reliably due to overlap of Sr (La1 – 1.81 keV, La2 – 1.80 keV, Lb1 – 1.87 keV) with Si (Ka1 – 1.74 keV, Ka2 -1.74 keV, Kb1 1.84 keV)

Some large crystals can be observed in the SEM, and coupled with the reduction of the O peak for these it would imply this could be some residual NaF/KF which was not washed out from the material, F cannot be readily observed due to the only Ka1 emission being at 0.677 keV where there is only a small peak which would likely not correspond to the significant decrease of the O peak. However, there could be more F present than indicated due to the possible reabsorption of the low energy F emission by heavier elements present.

For every strontium atom being incorporated into the structure via an ion exchange mechanism, two corresponding sodium or potassium atoms must be removed to maintain charge balance. To observe this another ion exchange experiment was performed, this time at a much lower concentration level. 50 ppm Sr(NO3)2 solution was used for the experiment to simulate a more realistic level of strontium in a waste stream. To avoid the problems encountered with XRF and

SEM/EDX, the ion exchange liquor was examined by ion chromatography, Figure 4.8, before

112 and after ion exchange to examine the composition.

SEPC2012new #30 Ca Start ECD_1 25.0 SEPC2012new #38 Na End ECD_1 µS 50.0 µS 7 - Calcium - 8.014 2 - sodium - 3.681 22.5 Strontium 45.0 Sodium

20.0 40.0

17.5 35.0

15.0

30.0

12.5

25.0

10.0 20.0 Strontium

7.5

15.0 5.0 Potassium

10.0

2.5

Signal Signal Sodium 4 - potassium - 4.854 3 - sodium - 3.677 6 - strontium1 - 7.214 7 - Calcium - 8.064 4 - potassium5 - -magnesuim 4.867 - 6.067 5.0 0.0 1 - 1.437 2 - 2.737

3 - 4.064 1 - 3.151 5 - magnesuim6 - strontium1 - 6.051 - 7.201 -2.5 0.0

min min -5.0 -5.0 0.0 1.3 2.5 3.8 5.0 6.3 7.5 8.8 11.0 0.0 1.3 2.5 3.8 5.0 6.3 7.5 8.8 11.0 Retention time (min) RetentionSodium time (min)

Figure 4.8 - Ion chromatogram of ion exchange solution before (left) and after (right) ion exchange with AV-7

The integrated intensities of the peaks, Table 4.5, directly correspond to the level of each ion in solution. It is evident the level of strontium is being significantly reduced and the sodium and potassium levels are indicative of an ion exchange reaction occurring. The Kd of the ion exchange shown in Figure 4.8 comes to ~1800.

Table 4. 5 - Integrated intensities and corresponding concentrations from ion chromatography

Peak Name Before exchange After Exchange Integrated Integrated intensity Concentration intensity Concentration (µS*min) (ppm) (µS*min) (ppm) Sodium 0.11 0.56 6.36 45.22 Potassium 0.03 0.29 1.04 10.99 Strontium 8.11 49.52 1.78 10.77

113

Having shown AV-7 does ion-exchange strontium over a range of concentrations, from 1.0 mol dm-3 to 50 ppm, further detailed ion exchange studies will be described in chapter 5.

4.3-5 Structural characterization

4.3-5a Time of flight neutron diffraction

TOF Neutron diffraction data were collected for the AV-7 system and the exchanged equivalent.

The data were calibrated using a NIST certified Si standard. Rietveld refinements (Tables 4.6-4.9 and Figures 4.9 & 4.10) were performed to obtain atomic positions and, in the case of the exchanged material, site occupancies. Refinement progressed by initially using structure reported by Lin et al.37, fixing framework positions and refining cation positions before allowing all atomic positions to refine. Constraints were initially used to keep water molecules in position, along with occupancy constraints to ensure charge balancing of shared Na/K occupancy sites, however, these constraints were not applied to the final cycles of the refinement. In the case of

Sr-exchanged AV-7 Sr positions and all cation occupancies were initially constrained to maintain charge balance and ensure Sr atoms remain in the pore; again these constraints were lifted for the final cycles of refinements. The data were collected at HRPD, ISIS running the samples for 10 hours and refined using TOPAS Academic 4.3 with a Chebyshev background function and a peak shape function for TOF neutron data contributed to the TOPAS Academic macro library by Professor W. I. F. David93.

114

Figure 4. 9- Refined structure of AV-7 with grey SnO6 octahedra, + blue SiO4 tetrahedra, purple K spheres and red/white structural water

115

Table 4. 6 - Refinement parameters for AV-7

a(Å) 6.47182(7) b (Å) 11.5540(1) c (Å) 12.9486(2) V (Å3) 935.07(2) β (°) 105.062(7)

space group P21/c Z 4 Nº independent parameters 82

Rwp 4.823

Rexp 1.602

Rp 4.224

Site x y z occ Beq Sn1 0.1135(6) 0.7232(3) 0.2192(3) 1 0.56(5) K1 0.685(1) 0.8796(7) 0.3918(6) 0.81(3) 1 Si1 0.8493(9) 0.4760(4) 0.1608(5) 1 0.23(4) Si2 0.2014(8) 0.7361(4) -0.0177(5) 1 0.23(4) Si3 0.5851(9) 0.6996(5) 0.1581(4) 1 0.23(4) Na1 0.792(1) 0.5473(8) 0.5908(6) 0.83 (6) 1 K2 0.839(6) 0.484(4) 0.593(3) 0.17(5) 1 O1 0.0650(6) 0.5508(4) 0.1916(3) 1 1.25(6) O2 0.0522(7) 0.7635(4) 0.0609(3) 1 1.25(3) O3 0.8022(6) 0.7660(3) 0.2077(4) 1 1.25(3) O4 0.1863(7) 0.8930(4) 0.2467(4) 1 1.25(3) O5 0.1544(7) 0.6822(4) 0.3771(4) 1 1.25(3) O6 0.4374(7) 0.6910(4) 0.2383(4) 1 1.25(3) O7 0.1551(7) 0.8992(4) 0.4450(4) 1 1.25(3) O8 0.3652(7) 0.0600(5) 0.3723(4) 1 1.25(3) O9 0.4548(7) 0.7489(4) 0.03972(4) 1 1.25(3) O10 0.6041(6) 0.5137(2) 0.3896(5) 1 1.25(3) D1 0.666(1) 0.4615(5) 0.3438(5) 1 4.62(9) D2 0.541(1) 0.5761(5) 0.3381(5) 1 4.62(9)

116

Table 4. 7 - Refinement parameters for Sr-exchanged AV-7

a(Å) 6.47000(5)

b (Å) 11.5495(1)

c (Å) 12.9458(7) V (Å3) 934.15(1) β (°) 105.0411(8)

space group P21/c Z 4 NO. independent parameters 85

Rwp 4.207

Rexp 1.657

Rp 3.071

Site x y z occ Beq Sn1 0.1134(8) 0.7202(4) 0.2161(4) 1 0.43(7) K1 0.746(5) 0.7528(2) 0.403(3) 0.25(1) 1 Sr1 0.691(2) 0.8842(9) 0.3924(9) 0.5(2) 1 Si1 0.8511(1) 0.4748(5) 0.1589(7) 1 0.50(6) Si2 0.203(1) 0.7354(5) -0.0180(6) 1 0.50(6) Si3 0.584(1) 0.7011(4) 0.1577(5) 1 0.50(6) Na1 0.688(9) 0.542(4) 0.650(4) 0.16(1) 1 K2 0.8443(5) 0.5370(4) 0.607(6) 0.239(8) 1 Sr2 0.776(5) 0.5458(2) 0.586(2) 0.30(8) 1 O1 0.0653(8) 0.5510(4) 0.1911(4) 1 1.37(4) O2 0.0535(8) 0.7634(4) 0.0626(4) 1 1.37(4) O3 0.8002(8) 0.7669(4) 0.2070(5) 1 1.37(4) O4 0.1811(9) 0.8918(5) 0.2456(4) 1 1.37(4) O5 0.1541(9) 0.6821(4) 0.3789(5) 1 1.37(4) O6 0.4299(8) 0.6929(5) 0.2371(5) 1 1.37(4) O7 0.1534(9) 0.9000(5) 0.4445(4) 1 1.37(4) O8 0.3634(6) 0.0624(5) 0.3715(5) 1 1.37(4) O9 0.4525(8) 0.7500(4) 0.03930(5) 1 1.37(4) O10 0.6068(8) 0.5151(3) 0.3923(4) 1 1.37(4) D1 0.6162(2) 0.4616(1) 0.3338(7) 1 4.78(9) D2 0.530(2) 0.5754(1) 0.3349(8) 1 4.78(9)

117

5 Observed Caclulated Difference 4

Counts 1

-1

-2 30000 40000 50000 60000 70000 80000 90000 100000 110000 120000 130000 TOF (mSec) 6 Observed Caclulated 5 Difference

Counts

-1

-2 30000 40000 50000 60000 70000 80000 90000 100000 110000 120000 130000 TOF (mSec)

Figure 4. 10 - Observed, calculated and difference powder neutron diffraction pattern of AV-7 (top) and Sr-exchanged AV-7 (bottom)

118

Table 4. 8 - Refined bond distances and valence sums from neutron TOF diffraction of AV-7 and Sr-exchanged AV-7

AV7 bond distances Sr-AV7 bond distances

Bond Distance Bond Bond Distance Valence Bond Distance Valence Bond Distance Valence Bond Distance Valence Valence Sn1_O1 1.971(5) 0.840 K1_O1 2.848(7) 0.126 Sn1_O1 1.991(5) 0.796 K1_O1 2.876(8) 0.117 Sr1_O1 2.876(8) 0.129 Sn1_O2 1.991(5) 0.797 K1_O2 3.190(8) 0.056 Sn1_O2 2.007(8) 0.762 K1_O2 3.234(4) 0.050 Sr1_O2 3.234(4) 0.049 Sn1_O3 2.015(5) 0.748 K1_O3 2.948(7) 0.099 Sn1_O3 2.045(3) 0.691 K1_O3 2.993(8) 0.089 Sr1_O3 2.993(8) 0.094 Sn1_O4 1.994(6) 0.791 K1_O4 3.169(7) 0.059 Sn1_O4 2.011(2) 0.756 K1_O4 3.222(2) 0.052 Sr1_O4 3.222(2) 0.051 Sn1_O5 1.988(5) 0.803 K1_O6 3.028(8) 0.082 Sn1_O5 2.021(3) 0.736 K1_O6 3.079(1) 0.073 Sr1_O6 3.079(5) 0.074 Sn1_O6 2.012(5) 0.754 K1_O7 2.907(8) 0.109 Sn1_O6 2.022(9) 0.733 K1_O7 2.925(9) 0.105 Sr1_O7 2.925(9) 0.113 Sum 4.733 K1_O8 2.801(7) 0.141 Sum 4.472 K1_O8 2.833(8) 0.130 Sr1_O8 2.833(9) 0.144 Si1_O1 1.536(9) 1.226 K1_O9 2.99(7) 0.090 Si1_O1 1.557(6) 1.166 K1_O9 3.124(8) 0.082 Sr1_O9 3.124(8) 0.066 Si1_O4 1.545(6) 1.201 Sum 0.762 Si1_O4 1.573(3) 1.124 Sum 0.699 Sr1_O1 Sum 0.796 Si1_O7 1.598(6) 1.062 Si1_O7 1.606(3) 1.042 Si1_O8 1.626(6) 0.995 K2_O1 2.857(3) 0.123 Si1_O8 1.652(5) 0.936 K2_O1 2. 876(9) 0.117 2.876(9) 0.129 Sum 4.484 K2_O2 2.824(3) 0.133 Sum 4.269 K2_O2 2.860(9) 0.122 Sr2_O1 2.860(7) 0.134 Si2_O2 1.549(6) 1.190 K2_O3 2.584(6) 0.235 Si2_O2 1.557(2) 1.167 K2_O3 2.611(1) 0.220 Sr2_O2 2.611(1) 0.264 Si2_O5 1.577(6) 1.115 K2_O4 2.855(5) 0.124 Si2_O5 1.592(4) 1.076 K2_O4 2.871(1) 0.119 Sr2_O3 2.870(8) 0.131 Si2_O7 1.614(6) 1.023 K2_O5 2.592(3) 0.231 Si2_O7 1.641(6) 0.960 K2_O5 2.611(8) 0.220 Sr2_O4 2.611(7) 0.263 Si2_O9 1.594(6) 1.072 K2_O9 3.080(5) 0.073 Si2_O9 1.604(2) 1.047 K2_O9 3.128(1) 0.065 Sr2_O5 3.128(3) 0.065 Sum 4.400 K2_O10 2.642(5) 0.205 Sum 4.250 K2_O10 2.552(4) 0.253 Sr2_O9 2.552(4) 0.309 Si3_O3 1.530(6) 1.243 Sum 1.123 Si3_O3 1.542(2) 1.209 Sum 1.117 Sr2_O10 Sum 1.295 Si3_O6 1.554(7) 1.176 Si3_O6 1.585(8) 1.093 Si3_O8 1.624(6) 1.000 Si3_O8 1.646(7) 0.949 Si3_O9 1.61(8) 1.033 Si3_O9 1.618(7) 1.012 Sum 4.452 Sum 4.262

119

Table 4. 9 - Refined bond angles from neutron TOF diffraction of AV-7 and Sr-exchanged AV-7

AV7 Bond angles Sr AV7 Bond angles

O1_Sn1_O2 94.67(2) O1_Si1_O4 116.3(4) O1_Sn1_O2 95.103(1) O1_Si1_O4 115.548(1) O1_Sn1_O3 97.19(2) O1_Si1_O7 109.06(3) O1_Sn1_O3 97.269(1) O1_Si1_O7 109.367(1) O1_Sn1_O4 175.88(2) O1_Si1_O8 111.07(3) O1_Sn1_O4 175.817(1) O1_Si1_O8 110.876(1) O1_Sn1_O5 85.78(2) O4_Si1_O7 108.96(3) O1_Sn1_O5 85.638(1) O4_Si1_O7 108.937(1) O1_Sn1_O6 88.27(2) O4_Si1_O8 106.47(3) O1_Sn1_O6 88.318(1) O4_Si1_O8 106.477(1) O2_Sn1_O3 86.60(2) O7_Si1_O8 104.25(3) O2_Sn1_O3 86.931(1) O7_Si1_O8 105.093(1) O2_Sn1_O4 85.74(2) O2_Si2_O5 114.11(3) O2_Sn1_O4 85.575(1) O2_Si2_O5 114.216(1) O2_Sn1_O5 175.87(2) O2_Si2_O7 105.94(3) O2_Sn1_O5 176.436(1) O2_Si2_O7 105.831(1) O2_Sn1_O6 95.25(2) O2_Si2_O9 113.48(4) O2_Sn1_O6 94.936(1) O2_Si2_O9 113.649(1) O3_Sn1_O4 86.92(2) O5_Si2_O7 109.05(3) O3_Sn1_O4 86.886(1) O5_Si2_O7 108.479(1) O3_Sn1_O5 89.26(2) O5_Si2_O9 106.98(3) O3_Sn1_O5 89.517(1) O5_Si2_O9 106.995(1) O3_Sn1_O6 174.09(2) O7_Si2_O9 107.02(3) O3_Sn1_O6 173.941(1) O7_Si2_O9 107.381(1) O4_Sn1_O5 94.11(2) O3_Si3_O6 111.86(3) O4_Sn1_O5 93.939(1) O3_Si3_O6 112.009(1) O4_Sn1_O6 87.62(2) O3_Si3_O8 111.11(3) O4_Sn1_O6 87.511(1) O3_Si3_O8 110.688(1) O5_Sn1_O6 88.87(2) O3_Si3_O9 112.03(3) O5_Sn1_O6 88.568(1) O3_Si3_O9 112.073(1) O6_Si3_O8 108.28(3) O6_Si3_O8 107.753(1) O6_Si3_O9 112.04(3) O6_Si3_O9 112.597(1) O8_Si3_O9 100.96(3) O8_Si3_O9 101.112(1)

120

The structure described by the refinement, Figure 4.10, shows mostly good agreement with the structure described in literature with overall slightly decreased Sn-O bond distances, e.g

Sn-O1 1.971(8) Å cf 2.084(8) Å. The Sn-O octahedron is also intact with a small deviation in bond angles, eg O1-Sn-O4 177.2(6) ° reported by Lin et al. with neutron refinement yielding the same angle as 175.88(2)°. Comparing the bond distances in Table 4.8 to the sum of the

Shannon radii of the atoms would suggest many of the bond distances are too small. Sn-O and Si-O bonds would be 2.19 Å and 1.64 Å, respectively. The bond distances observed are all smaller, however, this is not an uncommon observation in these types of materials49,60,90.

The K-O distance expected would be ~2.9 Å, however, sodium atoms also occupy the cation positions which would have an expected bond distance of ~2.56 Å. The average cation- oxygen bond distance in the cation sites in the unexchanged AV-7 are closer to the 2.9 Å K-O bond distance in the case of the K1 site and closer to the 2.56 Å Na-O bond distance in the case of the K2 site, this would suggest that the K2 cations sites are shared Na/K sites, which is also what is reported by Lin et al. 67. After exchange there is not a significant change in the average cation-oxygen bond distances, but there is a slight decrease as seen by a decrease in the bond valence sum from 0.762 to 0.699 for the K1 site and 1.123 to 1.117 in the K2 site.

As the Sr-O expected bond distance would be ~2.59 Å, any exchange would shift the average cation position closer to the framework decreasing the average bond distances.

121

Figure 4. 11 - Local Structure around cation exchange sites in AV-7 (left) and Sr-AV-7 (right) shown in relation to overall framework (below)

122

Whilst the original paper describes cation sites as one site containing solely K+ (K1) and a mixed occupancy site containing Na+ and K+ (K2), due to the similarity in the scattering cross sections of these elements in the case of neutrons, differentiation has proven difficult and for the sake of refinement only K+ has been used initially, adding in Na atoms at a later date.

Upon exchange no great structural change is seen to occur, however an increase in the unit cell volume is seen, along with an increase in the Si-O bond distances as seen in Tables 4.8 &

4.9. Whilst Sr2+ has a smaller ionic radius than K+, it is larger than Na+ and an expansion of the unit cell would be expected if a large amount of Na+ was replaced for Sr2+.

Figure 4.11 shows the immediate environment around the exchanged cation environment and water molecules.

Water molecules (W1) in the pore can be seen to have oriented in a manner so as to favourably align the lone pairs with the cations and undergo hydrogen bonding with the surrounding framework. The deuterium atoms D1 & D2 are within 1.8 Å from O4 and O6 respectively, as shown in Figure 4.11, indicating likely H-bonds. Upon exchange the angle between the neighbouring cations and water molecule increases from 97.21(8) ° to 99.13(7) °. This is likely due to the higher charge density of Sr2+ in comparison to Na+ and K+.

From the diffraction data it can be inferred Sr2+ exchanges into both sites with a slight preference for the K2 site. This preference can be rationalized using bond valence calculations, shown in Tables 4.8 & 4.9, which also indicate the larger, more accessible K2 site to be the preferred ion exchange site.

123

4.3-5b Temperature stability and variable temperature X-ray diffraction

According to Lin et al. 62 AV-7 can be thermally decomposed to a rhombohedral phase known as AV-11. To examine this expected phase transition TGA and variable temperature lab X-ray diffraction data were collected along on AV-7 and Sr-exchanged AV-7. High quality synchrotron data from the I-11 beamline at the Diamond Light Source were also collected on thermally treated samples of Sr-exchanged AV-7. Thermal gravimetric analysis of both pristine and exchanged AV-7, shown in Figure 4.12, appears to indicate a difference in the loss of structural water. A mass loss of ~4.8% over a temperature range between 30 and 800°C for AV-7, which agrees with that reported by Lin and co-workers, beginning at ~

200°C is seen to be more gradual with a more intense negative signal from the differential thermal analysis (DTA) indicating a more endothermic process. In the case of the Sr- exchanged AV-7 the mass loss is more rapid and the DTA peak less intense suggesting less tightly bound structural water.

Upon ion exchanging sodium for strontium, some of the pores will be less occupied due to removing two sodium atoms which are replaced by one strontium to maintain charge balance and creating pores with less cations. Removing the lone-pair interaction between the cation and water molecule could result in less tightly bound water and explain the trend observed in the TGA. Exchanging one Sr2+ cation in to the structure per two Na+ cations would result in a higher void volume in the pores of the structure and thus water in these empty pores would be less strongly bound, which agrees with the observation of water being removed slightly more easily after strontium ion-exchange.

124

AV-7

1000 102 0.0 101 800 -0.2 100 -0.4 600 -0.6 99

400 -0.8 98

exo

Temp/C -1.0 Mass/% 97 200 -1.2 96 -1.4 0 95 -1.6

-1.8 94 0 20 40 60 80 100 120 140 160 Time/min

Time/min vs Temp/C Time/min vs DTA/(uV/mg) Time/min vs Mass/%

Sr AV-7

1000 102

800 0.0 100

600 -0.2 98

400 -0.4

exo

Mass/% Temp/C 96 200 -0.6

94 0 -0.8

-1.0 92 0 20 40 60 80 100 120 140 160 Time/min

Time/min vs Temp/C Time/min vs DTA/(uV/mg) Time/min vs Mass/%

Figure 4. 12 - TGA trace of AV-7 (top) & Sr- exchanged AV-7 (bottom) heated under N2

125

70000

60000

50000

40000

Counts 30000 1000°C

900°C 20000 800°C 10000 30°C 0 10 15 20 25 30 35 40 45 50 (º) 2Theta Figure 4. 13 - Variable temperature PXRD of AV-7

Upon heating in air AV-7 does appear to undergo the phase change to AV-11, Figure 4.16,

90 R3h (NaK3SnSi6O18 ) via a dehydrated intermediate, based on indexing data collected from

lab X-ray diffractometers. Upon refining I11 data of the material collected ex-situ, Figure

4.13 and Table 4.10, a good fit was difficult to obtain and upon closer inspection it is evident

there are some cases of peak splitting throughout the pattern as shown in Figure 4.15 where

the (104) peak is split. This would indicate the structure may in fact be slightly deformed to

be monoclinic rather than rhombohedral.

180000 Observed Caclulated 160000 Calculated Difference 140000

120000

100000

80000

Counts 60000

40000

20000

-20000

-40000 0 10 20 30 40 50 60 70 80 2 theta (º)

Figure 4. 14 - Synchrotron PXRD data collected on AV-7 heated to 1100°C.

126

25000 Observed CalculatedCaclulated [TypeDifference a 20000 quote

15000 from the

10000 document

Intensity arb.u 5000 or the

0 summary

of an -5000 13.8 13.85 13.9 13.95 14 14.05 2 Theta (º) interesting

point. You Figure 4. 15 - Observed peak splitting from I11 data collection on heated AV-7

can

Table 4. 10 - Refinement information for synchrotron data collection of heatedposition AV-7

a (Å) 10.120(2) Z 9 the text b (Å) 10.117(1) Nº. independent parameters 104

c (Å) 14.768(2) Rwp 12.347 box

α (°) 90 Rexp 1.126

β (°) 90 Rp 9.146 anywhere γ (°) 120 V(Å3) 1309.45(4) in the Space group R3H document.

Site x y z Occ Beq Sn1 2/3 1/3 0.089(2) 1 Use1.188(1) the Sn2 0 0 0.237(2) 1/3 1.72(4) Si1 0.340(1) 0.067(1) 0.1813(9) 1 Drawing1.297(9) Si2 0.0039(2) 0.452(1) 0.015(2) 1 1.215(1) K1 1/3 2/3 0.180(1) 1.018(8) Tools1 tab K2 0.440(1) 0.333(1) 0.352(1) 0.602(1) 1 Na1 1/3 2/3 0.179(2) 1.031(8) to1 change Na2 0.430(1) 0.267(1) 0.412(2) 0.490(3) 1 O1 0.101(3) 0.171(1) 0.321(1) 1 the1.518(2) O2 0.739(2) 0.437(1) 0.164(2) 1 1.839(2) O3 0.267(3) 0.467(2) 0.342(2) 1 formatting1.485(4) O4 -0.008(3) 0.344(2) 0.387(1) 1 1.152(4) O5 0.296(2) -0.098(2) 0.237(2) 1 of1.212(3) the pull O6 0.204(1) 0.075(2) 0.163(1) 1 1.116(5) quote text

box.]

127

Table 4. 11 - Refined bond distances of heated AV-7 phase

Bond Distance(Å) Bond Distance(Å) Sn1 - O2 2.0456(3) K1 - O4 2.750(3) Sn1 - O3 2.0788(2) K1 - O1 2.868(4)

Sn2 - O6 2.0311(1) K2 - O1 2.771(2) Sn2 - O1 2.0354(2) K2 - O3 2.773(4)

K2 - O6 2.798(5) Si1 - O5 1.598(3) K2 - O5 2.825(2) Si1 - O6 1.603(3) K2 - O5 3.049(2) Si1 - O2 1.6145(4) Si1 - O4 1.6212(2)

Si2 - O3 1.596(2) Si2 - O1 1.610(3) Si2 - O5 1.631(2) Si2 - O4 1.639(4)

+ Figure 4. 16- Refined Structure of heated AV-7 phase with grey SnO6 octahedra, blue SiO4 tetrahedra and purple K spheres

128

This more dense form could potentially be used as a waste form as the smaller pores which contain no water molecules may be able to lock in exchanged cations and prevent back exchange. In order to examine the suitability for this material as a waste form AV-7 was Sr-

-3 exchanged with 1.0 mol dm Sr(NO3)2 for 24 hours then loaded into a can for hot isostatic pressing. The can was then cut into slices and the steel dissolved resulting in slices of a dense ceramic monolith. A standard method, MCC-ASTM, was then used to measure the leaching rates of the new material which will be discussed in Chapter 5.

4.4 Conclusions

Brief investigations into preparing new stannosilicate materials have been performed with one new material prepared and the structure solution underway. The synthesis of tin silicates can be challenging due to the harsh conditions necessary to prevent formation of the stable

SnO2 phase. Tin kostylvite has been prepared as per in literature and the structural changes upon strontium ion exchange and heating investigated, using laboratory X-ray powder diffraction and neutron powder diffraction. Neutron powder diffraction shows water molecules in the large pore of the structure coordinating with both the framework and cations within the pore, possibly allowing for extra stabilisation of the cations.

129

Chapter 5: Ion exchange

5.1 Introduction

This chapter describes in detail more extensive ion exchange studies of the materials discussed previously; the initial studies in Chapters 3 and 4 were performed using higher concentrations to screen for potential ion exchange and to examine the maximum uptake of materials and were based on X-ray fluorescence analysis. Following successful higher concentration ion exchange, low level (1-100 ppm) exchanges were performed and monitored using ion chromatography, which also allows for monitoring the increase in solution of the cations exchanged out of the framework in addition to the uptake of the target cations. Finally the lowest level exchanges, which are at more realistic concentrations for nuclear waste streams, were monitored using exchange solutions containing 85Sr measured using scintillation counting of the emitted gamma rays. Calcium and caesium were chosen as competing cations amongst others, as calcium is known to compete for exchange sites with strontium due to their similarity in size and charge and caesium is commonly found in the same effluent streams along with strontium.

AV-3 and AV-7 were compared with a range of materials over high concentration: a small pore zeolite ETS-4 (Na9Si12Ti5O38(OH)·xH2O), a natural mineral clinoptilolite used in

SIXEP, and NaZrSi2O7, a material studied for its ion exchange properties during a related 4th year MSci project. Also included is a sample of AV-7 heated to 1050°C until transformed to a more condensed rhombohedral structure, AV-11.

130

5.2 Experimental

5.2-1 AV-7 Competitive strontium ion exchange

1.0g of AV-7 solid was added to 20cm3 of:

-3 a) 0.5 mol dm Sr(NO3)2 solution;

-3 -3 b) 0.5 mol dm Sr(NO3)2 with 0.5 mol dm Ca(NO3)2 solution; or

-3 -3 c) 0.5 mol dm Sr(NO3)2 with 0.5 mol dm CsNO3 solution.

Solutions were shaken overnight for 24 hours before using centrifugation to separate the solid. The solid was then washed twice with portions of deionised water then centrifuged again. The washed solid was then formed into fused beads using the method described in

Chapter 2 for XRF analysis.

5.2-2 Ion Chromatography

In order to calibrate the ion chromatography column certified standard solutions were purchased from Fisher of 1000ppm for each element to be analysed. From these standard solutions were made 8 solutions of varying concentrations which were run in triplicate after flushing the system with water filtered through a millipore water filtration system. Plotting the integrated peak intensities vs the concentration gave a linear fit and allowed for calibration using the equation of the line from a linear regression, Figure 5.1 and Table 5.1.

131

50 Sodium 40 Potassium

PeakArea Magnesium 30 Calcium 20 Strontium

0 0 50 100 150 200 250 Concenration (ppm)

Figure 5. 1 - Calibration curves for ion chromatography column (error bars too small to be seen)

Table 5. 1 – Linear fits for ion chromatography calibrations

Calibration Linear Fit standard Sodium y=0.2496x+7e-15 Potassium y=0.1608x-7e-15 Magnesium y=0.3711x Calcium y=0.1419x+3e-15 Strontium y=0.3007x

0.2 g of each ion exchange material was added to 100 ppm solutions of Sr(NO3)2 to examine low level ion exchange properties, chromatograms of the ion exchange solution before and after ion exchange allowed for monitoring of both uptake and release of cations.

5.2-3 Initial 85Sr exchanges

1.00 g of each ion exchange material was batch exchanged with 10 cm3 of a 0.5 mol dm-3

85 solution of Sr(NO3)2 doped with Sr over 24 hours. The Sr(NO3)2 solution was prepared by

3 dissolving 10.582 g Sr(NO3)2 in 100 cm of deionized water followed by the addition of ~15

132

85 MBq of Sr in the form of SrCl2 in water. The solid was separated from the supernatant, washed with water and dried for analysis using a Ge crystal scintillation counter.

5.2-4 Trace 85Sr ion exchanges pH 7 & 4

Following initial ion exchanges a second and third batch of experiments were performed to look at exchange over a range of materials over four sets of conditions.

0.05 g of each ion exchanger was added to 10 cm3 of solution as listed in table 5.2. The suspensions were shaken overnight for 24 hours before centrifuging to separate the solid and supernatant fluid. The supernatant was collected and the solid washed with deionised water before centrifuging again. Both the solid and supernatant fluid were counted using a Ge- based liquid nitrogen cooled scintillation counter to measure the activity. The integrated intensities of the peaks generated by the MAESTRO software were used for the activity of

85Sr in each solution as there should be a linear relationship between the two and only relative numbers rather than absolute are necessary.

133

Table 5. 2 - Composition of ion exchange liquors - 85Sr concentration given in MBq due to a very low chemical concentration

85 SrCl2 Sr(NO3)2 KNO3 NaNO3 Mg(NO3)2 Ca(NO3)2 CsNO3 HCl pH

Sr pH 4 0.22 MBq 10 ppm ~5µl 4.1 (<1 ppb)

Competitive 0.22 MBq 10 ppm 10 ppm 100 ppm 20 ppm 20 ppm 10 ppm 6.9 pH 7 (<1 ppb)

Sr pH 7 0.22 MBq 10 ppm 6.8 (<1 ppb)

Competitive 0.22 MBq 10 ppm 10 ppm 100 ppm 20 ppm 20 ppm 10 ppm ~5µl 3.9 pH 4 (<1 ppb)

134

5.2-5 Trace 85Sr ion exchanges pH 10-11

0.01g of ion exchange material was added to a centrifuge tube followed by the addition of 10 cm3 of ion exchange solution A, B, C or D which had the compositions shown in Table 5.3.

When adding strontium to each of the solutions, non-radioactive strontium was added first to each of the solutions so the addition of equal portions of 85Sr to each solution brings the total strontium concentration up to the value tabulated. Samples were prepared in triplicate for a total of 12 ion exchanger-solution samples for each examined ion exchange material. These ion exchanges were left to shake overnight for 24 hours before using centrifugation to separate the solids from the liquids.

Table 5. 3 - Composition of ion exchange solutions A-D in ppm

Solution A B C D pH 11.2 11 10 10 Cs (ppm) 9.06x10-4 0.362 0.302 1.51 85Sr + Sr (ppm) 1.51×10-5 1.53×10-3 5.67 ×10-3 3.21 ×10-2 Na (ppm) 70 130 60 210 K (ppm) 0.2 10 20 250 Ca (ppm) 0.2 3 1.2 60 Mg (ppm) 0.1 3 140 30

In order to counter previously encountered problems during the pH 7 & 4 85Sr exchanges, where activity from all samples in the room resulted in a high background count, the method for this batch of exchanges was modified. A new detector was purchased to allow us to modify the instrument set-up more freely. The new detector uses a NaI scintillation crystal rather than pure-Ge, which whilst less efficient does not require liquid nitrogen cooling. In addition large amounts of lead flashing were used as shielding around both the detector and other samples in the area to reduce the background counts by a factor of 10. In order to utilize the activity of both the solid and liquid elements of the samples, solutions of known activity

135 were used to calibrate the detector. In the case of calibrating for the liquid samples aliquots of known 85Sr activity were added to 10cm3 and for calibrating the solid samples, aliquots between 20 and 200 µl of higher activity 85Sr were pipetted into the bottom of the centrifuge tubes used for measuring, to better simulate the activity from a point source.

5.3 Results and discussion

5.3-1 AV-7 Competitive strontium ion exchange

Initially the main focus of the project was the investigation of the ion exchange properties of

AV-7 (NaKSnSi3O9·H2O), a synthetic tin form of the natural zirconosilicate kostelyvite, reported by Lin et al. 60. Preliminary ion-exchange results are described in Chapter 3, which were followed by a brief investigation into the competitive strontium exchange.

Table 5. 4 - Mole ratio compositions of solid AV-7 after ion exchange, XRF data normalized on Sn content

Sn Si Na K Cs Sr Unexchanged NaKSnSi3O9 1.00(4) 2.8(2) 1.6(3) 2.91(4) 0.00 0.00 0.5M Sr2+ with 0.5M Ca2+ 1.00(4) 3.2(2) 1.2(3) 2.18(4) 0.00 0.38(3) 0.5M Sr2+ with 0.5M Cs+ 1.00(4) 3.3(2) 1.1(3) 2.13(4) 0.07(5) 0.94(3) 0.5M Sr2+ with 0.5M Ca2+ & 0.5M Cs+ 1.00(4) 3.3(2) 1.7(3) 2.16(4) 0.08(5) 0.31(3) 0.5M Sr2+ only 1.00(4) 2.7(2) 1.1(3) 1.58(4) 0.00 0.80(3)

The measurements were taken by forming a fused bead using the method described earlier in

Chapter 2, whereby the solids remaining after ion exchange reactions are filtered, washed and dried before being heated to 1050°C in a lithium borate glass to form a homogenous glass bead. The XRF measurement method used a calibrated method with a number of fused beads

136 of standard composition to analyse only Sn, Si, Na, K, Cs and Sr and thus any Ca which would have been exchanged into the material is not visible. A limitation to this method is the time-consuming nature of the sample preparation and time taken to calibrate the instrument as any element not included in the initial calibration will not be measured. In addition to a missing element, in this case calcium, not being measured, due to the matrix effect equation as discussed in Chapter 2, if calcium forms a significant missing mass fraction then the attempt of the spectrometer software to assign a mass fraction based upon the observed fluorescence will be incorrect, however, this effect is usually minimal unless the missing element forms a substantial fraction of the sample. Some strontium exchange is noticeable despite the presence of Ca2+ and Cs+, with Ca2+ affecting uptake more than Cs+. Whilst sensitivity to Ca2+ is not desired, it is not surprising as strontium and calcium are challenging to separate. However, the apparent preference of strontium over caesium is a useful feature as many of the ion exchangers in use such as Nb-CST and clinoptilolite are selective for caesium over strontium.

5.3-2 Ion Chromatography

Ion chromatography was used initially to measure the Sr2+ concentrations of the ion exchange solution before and after ion exchange. As shown in Figure 5.2, this allows for a comparison of AV-3 and AV-7 with clinoptilolite, a zeolite known to have good ion exchange behaviour and titanite, a previously prepared material which may have potential ion exchange behaviour. These measurements were performed for some initial results to test the viability of the ion chromatography because of the rapid throughput available as opposed to using another technique such as inductively coupled plasma optical emission spectrophotometry

(ICP-OES) or atomic absorption spectrometry which would have required a long waiting time for a high number of samples.

137

900

800

700

600

1 - 500 pH 7 400

pH 11 Kd cm3 Kd g 300

200

100

0 ClinoptiloliteClinoptiloite PerarasiteAV-3 Tin KostyleviteAV-7 Titanite

Figure 5. 2 - Material comparison for 100 ppm Sr ion exchange as measured by ion chromatography

Figure 5.2 indicates the relative performance of these materials, with clinoptilolite clearly showing the highest uptake, however, it is interesting to note that clinoptilolite shows reduced uptake with increased pH whereas the opposite is seen with AV-3, AV-7 and titanite. AV-3 appears to have the best performance of the new materials, possibly due to the higher surface area in comparison to the other materials due to the crystals being formed of clusters of thin sheets as shown previously through SEM images in Chapter 3. Whilst a higher surface area would not affect the bulk ion exchange properties if the system has reached equilibrium, it is possible that 24 hours is not sufficient time for equilibrium to be reached which would result in the ion exchange kinetics having a stronger influence over the observed strontium uptake.

Analysis of the chromatograms for the exchange of AV-7 in Figure 5.3 and also shown in

Chapter 4 in Figure 4.8 and Table 4.5, show the strontium peak clearly decreases and a sharp increase in sodium and potassium indicates the ion exchange process, where strontium enters the pore and sodium and potassium are removed from the pores, according to the integrated intensity of the peaks the strontium concentration is reduced from 105 ppm to 24 ppm and an

138 increase in sodium and potassium from trace amounts to 45 ppm and 11 ppm, respectively.

-6 -5 This corresponds to 5.4 × 10 moles of strontium into the framework and 1.3 × 10 moles of sodium and potassium combined out of the framework which is concurrent with the ion exchange expected, approximately twice the amount of monovalent cations released compared to divalent Sr2+removed from solution.

SEPC2012new #30 Ca Start ECD_1 SEPC2012new #38 Na End ECD_1 25.0 50.0 µS µS

7 - Calcium - 8.014 2 - sodium - 3.681 22.5 Strontium 45.0 Sodium

20.0 40.0

17.5 35.0

15.0 30.0

12.5 25.0

10.0 20.0 Strontium

7.5

15.0 5.0 Potassium

10.0

2.5

Signal Signal 4 - potassium - 4.854 3 - sodium - 3.677 6 - strontium1 - 7.214 7 - Calcium - 8.064 4 - potassium5 - -magnesuim 4.867 - 6.067 5.0 0.0 1 - 1.437 2 - 2.737

3 - 4.064 1 - 3.151 5 - magnesuim6 - strontium1 - 6.051 - 7.201 -2.5 0.0

min min -5.0 -5.0 0.0 1.3 2.5 3.8 5.0 6.3 7.5 8.8 11.0 0.0 1.3 2.5 3.8 5.0 6.3 7.5 8.8 11.0 Retention time (min) Retention time (min)

Figure 5. 3 - Example chromatogram for AV-7 before (left) and after (right) ion exchange

The chromatograms of AV-3, Figure 5.4 and Table 5.5 indicate the strontium concentration drops from 133 ppm to 89 ppm and sodium increases from trace amounts to 62 ppm corresponding to a strontium uptake of 3.01 × 10-6 moles and the release of 6.02 × 10-5 moles of sodium, approximately two and half times the expected amount of sodium released.

139

Sodium Strontium

Signal Strontium Signal

Retention time (min) Retention time (min)

Figure 5.4 - Example chromatogram for AV-3 before (left) and after (right) ion exchange

Table 5. 5 - Integrated intensities and concentrationof AV-3 Ion exchange liquor as measured by ion chromatography

Peak Before exchange After Exchange Name Integrated intensity Concentration Integrated intensity Concentration

(µS*min) (ppm) (µS*min) (ppm) Sodium 0.04 0.31 12.00 62.82 Strontium 14.45 133.78 9.43 87.34

5.3-2a Initial exchanges

These exchanges were repeated, using the same 100 ppm Sr(NO3)2 concentrations but

modified to contain 100 ppm of a competing cation. Examining the uptake shown in Figure

5.5, only calcium appears to have a significant detrimental effect on the uptake. The ionic

radii of Ca2+ and Sr2+ are reasonably similar ranging from ~100-134 pm and 118-144 pm

respectively, depending on coordination number. The similar size and charge and both being

group II elements result in very similar chemical behaviour meaning it would not be

surprising for calcium to strongly compete for the same cation site during ion exchange.

140

300

250

200

1 -

150 Kd cm3 Kd g 100

0 Na K Mg Ca Sr

Figure 5. 5 - Batch distribution coefficients for competitive 100 ppm Sr ion exchange of AV-3 as measured by ion chromatography

5.3-3 85Sr ion exchanges

5.3-3a Initial exchanges

Initial radioisotope ion exchange experiments were performed on AV-7, AV-3 and titanite along with two well-known materials for comparison, clinoptilolite and niobium-doped crystalline silicotitanite. These exchanges were performed under slightly acidic conditions, ~ pH 6 due to the acidity of the water supply and strontium nitrate in solution.

As seen in fig 5.6 AV-7 exhibited a relatively high Sr uptake in comparison to both AV-3 and titanite, however, both clinoptilolite and Nb-CST show higher uptake.

141

20000 18000 16000

14000

) 1

- 12000

g 3 10000

8000 Kd (cm Kd 6000 4000 2000 0 AV7 Titanite AV3 Clino Nb-CST

Figure 5. 6 - Batch distribution coefficients for initial 100 ppm 85Sr ion exchange experiments

It is evident the strontium uptake of titanite is poor despite the porous structure and similarity to other materials. There are a number of potential reasons for the lack of ion exchange:

 Lack of water in the pores; unlike the other materials titanite is not a

hydrated phase as there are no water molecules within the pores

indicated by XRD. It is possible that water molecules within the

channels of a material aid in the exchange mechanism in some way,

such as solvating the outgoing cation to stabilize it. This trend is also

noticeable in fresnoite, another porous anhydrous material which

shows poor ion exchange properties;

 The pore size of titanite is too low; at the narrowest point, from the

corner of one octahedron to the opposite tetrahedron, the pores are only

3.1 A wide, as shown in Figure 5.7, which may be too small to allow

the necessary ion mobility throughout the pore to enable exchange;

142

 The calcium cations are bound too tightly electrostatically for an ion

exchange to be favourable; in order to examine this possibility some

modelling using static energy calculations was intended, however, due

to difficulty finding correct potentials this work was not completed due

to time constraints.

In these initial experiments AV-7 is observed to perform better than AV-3, both of which are exchanging at a reasonable level comparable to clinoptilolite obtained from NNL and Nb- doped CST. However, it should be noted these results were obtained before the fine tuning of the measurement method, in particular the count time of the samples for these measurements was 1 minute of live time, in which the detector was occasionally saturated leading to high proportion of dead time (up to 45%). Later measurements set thr gamma detector live time to

20 minutes per sample which gave much less dead time (max 2%) a significantly better counting statistics.

143

Figure 5. 7 - Structure of titanite illustrating pore diameter

5.3-3b Trace 85Sr ion exchanges pH 7 & 4

These spiked exchanges were performed over two batches, using 100 ppm Sr(NO3)2 doped with 85Sr, first examining strontium uptake in an acidic environment with no competition and strontium uptake in a pH neutral environment with competing cations, followed by pH neutral with no competition, and acidic with competition. Ideally these exchanges would have been completed in triplicate, however, due to only having small amounts of 85Sr which had already decayed somewhat, it was not possible to complete such a large number of samples and still have sufficient counts necessary for reasonable counting statistics.

144

Many problems were encountered when measuring these samples. The geometry of a sample has a large impact on the counting, whilst all liquid and all solid samples should have similar geometries in relation to the detector the solid samples act more as a point source compared to the liquid samples where the 85Sr is distributed throughout a larger volume. This effect results in significantly higher counts for the solid samples than for the liquid samples, in fact many of the solid samples read higher than the original stock solution making it impossible to use all the data acquired. In hindsight, shaking the solid samples in water to disperse the particles may have helped to alleviate this problem, but due to long counting times much of the solid would have settled again during the measurement resulting in an incorrect reading.

Another solution to this problem and an improvement to the method in general would have been to calibrate the detector using stock 85Sr solution of a known activity, both in a dispersed solution and in a more concentrated space as in the solid samples. Not only would this have enabled the use of the activities measured from solid samples to be used for more than just relative checks between samples but also to ensure that the assumed linear relationship between the integrated area of the signal peak and the activity of the sample is indeed true.

Furthermore due to the high energy of the 85Sr emission (514 keV) and the high sensitivity of the detector, even with lead shielding around the detector background, counts from other samples in the area provided a significant background which had to be mathematically subtracted.

The exchange liquors used for comparison were based upon work performed by Dyer et al65 and the advice of Dr Zoe Maher of the NNL. In addition to AV-7 and AV-3, poorly crystalline zeolite with the gismondine (GIS) framework with an unknown Si/Al ratio prepared from coal-fly ash by R. Sommerville66 and nenadkevichite,

(Na,Ca)(Nb,Ti)Si2O7·2H2O, prepared by a MSc project student James Moore were investigated with pH neutral Sr exchange and acid competitive exchange. It should be noted

145 that a significant variable in this test is the surface area of the particles. All materials were hand ground in an agate pestle and mortar and it was not possible to ensure all materials have the same particle size and thus surface area.

4500 4000

3500

) 3000

- g

3 2500

cm 2000

Kd ( 1500 1000 500 0

Figure 5. 8 - Batch distributions for multiple materials ion exchanged with 100 ppm Sr only at pH 7

Neutral pH strontium exchange, Figure 5.8, was used as a baseline to compare the other exchange conditions, AV-7 does exhibit uptake but significantly lower than other materials which are shown to perform comparably to industrially used CST and clinoptilolite. Unlike in

85 -3 the initial Sr exchange, Figure 5.6, at a higher concentration of 0.5 mol dm Sr(NO3)2, AV-

3 is shown have higher uptake than AV-7. This could be due to a high degree of batch variation of the materials as the oven temperature can fluctuation greatly and small variations in the pressure of autoclaves due to varying degrees of leaking of the Teflon liners could have an effect on the synthesis.

146

3000

2500

2000

)

1 -

g 1500 3

1000 Kd (cm Kd

500

-500

Figure 5. 9 - Batch distributions for multiple materials ion exchanged with Sr only at pH 4

Comparing Figures 5.8 and 5.9, With a decrease in the pH and no competitive cations AV-7 and AV-3 are shown (Figure 5.9) to ion exchange to a lesser degree than both Nb- CST and clinoptilolite interestingly the decrease in AV-3 is more significant than the decrease in clinoptilolite which might be expected to show a reduction in ion exchange due to dealumination from the low pH, however, the high Si/Al ratio would likely stabilize the framework and prevent dealumination under the moderately acidic conditions used here.

Zeolite (GIS) retains a good uptake of strontium with the low pH actually improving the uptake of the latter.

Examining Figure 5.10 and comparing with Figure 5.8 indicates that whilst the strontium uptake of CST, AV3 and zeolite (GIS) and are all reduced upon the addition of competing cations, the uptake of AV-7 and clinoptilolite are mostly unaffected. The decreased uptake of

AV-3 is less significant in the case of addition of competing cations than the decrease upon acidifying the solution.

147

1200

1000

)

1 800

g 3 600

Kd (cm Kd 400

200

Figure 5. 10 - Batch distributions for multiple materials ion exchanged with Sr and competing cations at pH 7

3000

2500

2000

) )

g 3

1500 cm

Kd( 1000

500

Figure 5. 11 - Batch distributions for multiple materials ion exchanged with Sr and competing cations at pH 4

148

With both an acidic solution and the introduction of competing cations, Figure 5.11, all materials bar AV-7 show reduced uptake. Whilst AV-3 does not exhibit the highest uptake in any situation it does show consistently good uptake. Whilst low, the uptake of AV-7 appears to be mostly unaffected regardless of conditions. Surprisingly the uptake of CST and in fact many other materials, is higher under acidic conditions than under neutral conditions when competitive cations are present; this runs contrary to most literature, and so perhaps a direct comparison should not be made but rather comparing within the series. This anomaly would indicate that this series of measurements should be repeated with a number of replicate samples.

In comparison to the initial exchanges shown at the start of section 5.3-3, the Kd values obtained for AV-7 are substantially different. This may be due to differences in the batch of

AV-7 used or alternatively a dependence upon solid to solution ratio. Although theoretically the solid/solution ratio should be normalized by the definition of the Kd model, there may be some ion exchange sensitivity to concentration. A large solid/solution ratio would also result in solid particles coagulating together into larger particles which have a lower surface area and thus lessened adsorption on the surface. Further it may also be possible that the short period of 24 hours may not be sufficient to allow equilibrium to be reached, where exchange from the surface adsorption sites into the bulk exchange sites occurs, reducing the ion exchange uptake. These factors could potentially result in a reduced Kd measurement in systems with a high solid/solution ratio.

5.3-3c Trace 85Sr ion exchanges pH 10-11

Having encountered problems with the previous method of 85Sr ion exchanges, another set of exchanges were performed using a new scintillation counter to measure the activity of

149 samples and ion exchange solutions suggested by the NNL as described in the experimental section of this chapter and shown here in Table 5.6.

Table 5. 6 - Ion exchange solutions compositions in ppm

Solution A B C D pH 11.2 11 10 10 Cs (ppm) 9.06×10-4 0.362 0.302 1.51 85Sr + Sr (ppm) 1.51×10-5 1.53×10-3 5.67×10-3 3.21×10-2 Na (ppm) 70 130 60 210 K (ppm) 0.2 10 20 250 Ca (ppm) 0.2 3 1.2 60 Mg (ppm) 0.1 3 140 30

70000

60000

50000

)

1 -

g 40000 3 AV3

(Cm AV7

d d 30000 K Clino 20000 Nb-CST 10000

0 A B C D Exchange Solution

Figure 5. 12 - Comparison of batch distribution coefficients for strontium of multiple materials exchanged in a range of different solutions

As seen in Figure 5.12, the results for Nb-CST in solutions C where concentrations of magnesium are high & D with high concentrations of both sodium and potassium appear anomalous due to the Kd values being significantly higher than expected in comparison to solution A which should act as a baseline for comparison having no significant concentration of any competing cations. The very high uptake is observed is likely due to all the strontium

150 being removed from solution which results in a very high number. AV-3 demonstrates reasonable uptake, performing well in comparison to Nb-doped CST and better than clinoptilolite in solutions A and B where strontium is in higher concentrations relative to other elements in the ion exchange solution.. The reasoning behind the poor selectivity of

AV-3 is unclear as all of competing cations in the solutions have similar ionic radii between

100-150 pm which would mean a size-match effect is unlikely. The increased uptake of AV-3 over AV-7 and clinoptilolite may be due to the increased surface area of AV-3 as previously mentioned in this chapter, due to the morphology as seen in Chapter 3, where SEM images show the crystals are composed of clusters of parallel aligned plates which result in a higher surface area, however, this would only be in the case of equilibrium not being reached.

Overall high pH appears to reduce the influence of calcium and magnesium competition, possibly due to the formation of insoluble Ca(OH)2 and Mg(OH)2 compounds in suspension which are not readily ion exchanged.

5.3-4 Hipping and leach testing

The stability of a wasteform is important considering spent ion exchange material will likely be stored for decades; but the long chemical stability needs to be extrapolated from short term leach tests performed according to standard methods. Two standard methods are commonly used, MCC (C 1220) based on a dense monolith with a known surface area and PCT (C 1285) where a powder is used and the surface area is either measured using a technique such as nitrogen adsorption and a BET isotherm or geometrically estimated using the particle size.

In order to produce a dense monolith for the MCC test, a process known as HIPing (Hot

Isostatic Pressing) as previously described in Chapter 2 was used. 10 g each of AV-7 and

151

-1 3 AV-3 were exchanged in 0.05 mol dm Sr(NO3)2 to 1.02 wt% and 1.91 wt% strontium respectively before HIPing at 1100 °C.

The HIPing process for AV-7 produced a uniform black material, Figure 5.13, however X-ray diffraction, Figure 5.14, revealed the sample to contain FeO and some tin metal as the only observable crystalline phases, although some peaks may be obscured due to the iron content causing significant fluorescence in addition to the amorphous content. The reducing conditions of the HIP can has not only caused the tin in the sample to be reduced but also caused the iron from the steel can to be drawn into the material and contaminate the material making this sample brittle and mechanically fragile.

Figure 5. 13 - Image of AV-7 post HIPing having cut the HIP can to reveal a dense black solid

152

4000 Observed x

3500

3000

Counts 2500

2000 *

1500 10 20 30 40 50 60 2 Theta (°)

Figure 5. 14 - XRD pattern of HIPed AV-7 with peaks x - FeO * - indicating Sn0

In the case of AV-3 the HIPing caused the formation of several different phases, which can

be seen in the cross section of HIP can. The outer green edge indicates where the material

reacted with the iron in the steel again and the core can be seen to be inhomogeneous as

shown in Figure 5.15.

153

Figure 5. 15 Image of AV-3 post HIPing having cut the HIP can to reveal a dense green-white inhomogeneous solid

Powder X-ray diffraction, Figure 5.16, shows a number of different phases present, the main

phase being parakeldyshite (HxNa2-x Zr Si2O7) as described in Chapter 3.

35000 Observed CalculatedCaclulated 30000 [Type a quoteDifference

25000 from the 20000 document or 15000

Counts 10000 the summary

5000 of an

0 interesting -5000

-10000 point. You 10 20 30 40 50 60 2 Theta (°) can position

Figure 5. 16 - Refinement of HIPed AV-3 to identify parakeldyshite phase, seethe Chapter text box 3 for details

anywhere in 154 the document.

Use the

Drawing

Table 5. 7 - Leach testing water concentrations as measured by ICP-OES & ICP-MS * indicating anomalous result

Sr Sample K (mg/L) Zr (mg/L) Sn (mg/L) Cl (mg/L) Na (mg/L) Si (mg/L) (mg/L) ICP-OES ICP-MS ICP-MS ICP-MS ICP-MS ICP-OES ICP-OES Blank 0.05 8.1x10-4 < 2x10-4 1.16x10-3 0.324 0.39 0.242

AV-7 A- 15 7.8 2.93x10-2 - 1.73* - 2.8 68.9 days AV-7 B- 30 7.7 1.06x10-1 - 9x10-5 - 1.63 102 days (1) AV-7 C- 30 5.6 4.04x10-2 - 2.26x10-2 - 1.85 61.4 days (2) AV-7 D- 30 5.5 5.5x10-3 - 4.11x10-2 - 1.86 105 days (3)

AV-3 A- 15 - 1.24x10-2 0.506 - 48.5 80.9 47.1 days AV-3 B- 30 - 2.36x10-2 0.385 - 233 260 96 days (1)

AV-3 C- 30 - 1.62x10-2 0.296 - 207 250 113 days(2)

AV-3 D- 30 - 2.49x10-2 0.517 - 130 172 90.6 days(3)

Table 5.7 gives the concentrations of various elements obtained by ICP-MS and ICP OES

(performed by Robert Clough of Plymouth University) of the leachate solutions after

following leach testing procedure MCC (C 1220). Regrettably sample AV-7 A appears to

have been contaminated from another source of tin, possibly from a contaminate of the steel

HIP can which may not have been completely removed from the monolith prior to testing.

Also Sample AV-7B appears to have an unusually low concentration of tin, in this case lower

than the supplied blank so AV-7 A and AV-7 B have been ignored for the case of tin

leaching.

155

Table 5. 8 - Normalized elemental mass loss (g m-2)

Sample K Sr Zr Sn Cl Na Si

AV-7 A- 15 4.71 0.15 - 1.27 - 5.1 39.05 days AV-7 B- 30 4.66 0.55 - 0 - 2.62 57.88 days (1) AV-7 C- 30 3.37 0.21 - 0.02 - 3.09 34.79 days (2) AV-7 D- 30 3.3 0.02 - 0.03 - 3.11 59.59 days (3)

AV-3 A- 15 - 0.12 0.14 - 61.21 42.78 18.42 days AV-3 B- 30 - 0.24 0.1 - 295.65 137.94 37.64 days (1) AV-3 C- 30 - 0.16 0.08 - 262.61 136.63 44.32 days (2) AV-3 D- 30 - 0.26 0.14 - 164.77 91.18 35.52 days (3)

-2 Table 5.8 shows the normalized elemental mass loss (NL)i in g m which is given by:

(퐶푖푗−퐵푖)푉푗 (푁퐿)푖 = (5.1) 푓푖푆퐴 where,

Cij is the concentration of element i observed in the leachate from sample j

Bi is the concentration of element i observed in the blank solutions averaged over replicate blank solutions

Vj is the volume leachate added to sample j

156 fi is the mass fraction of element i in the unleached sample

SA is the surface area in m2

In this case the concentration in the samples is obtained by XRF of the sample after grinding and fusing into a bead shown in Table 5.9. The surface area was calculated using image processing software ImageJ to measure the pixels within the area of the samples from pictures taken of the samples next to a ruler for scale using a digital camera.

Table 5. 9 – Composition of HIPed AV-7 and AV-3 as measured by XRF

Sample K (wt%) Sr (wt%) Zr (wt%) Sn Cl (wt%) Na Si (wt%) Fe (wt%) (wt%) (wt%) HIPed 16.5(2) 1.90(3) - 13.60(2) - 4.7(3) 17.6(4) 44.21(2) AV-7 HIPed - 0.94(3) 37.15(2) - 2.9(5) 18.8(3) 23.4(4) 3.56(2) AV-3

The overall leaching observed for both AV-3 and AV-7 HIPed phases indicates a reasonably high leaching of silicon and sodium and potassium, which could indicate the formation of glassy sodium/potassium silicate phases which are more easily leached into solution.

Chloride still present in the AV-3 appears to leach very readily, considering the higher sodium leaching of AV-3 there may also be chloride in a sodium silicate glassy phase as there is no NaCl evident in the PXRD patterns of this material. Overall the cation leach rate of both

AV-3 and AV-7 appears to be significant, with the exception of strontium, however, this may simply be due to only small amount of strontium being present, and so a greater portion of the strontium present is in the bulk rather than the surface. Due to the high leaching rates and in the case of AV-7 poor mechanical properties, it would appear that the HIPing process under these conditions is not suitable for either AV-7 or AV-3.

157

5.4 Conclusions

This chapter contains work performed to study the ion exchange properties of AV-7, AV-3 and titanite in reference to other known ion exchange materials clinoptilolite and Nb-CST.

Ion exchange experiments were performed over a range of conditions, with variation of strontium concentration, competing cation concentration and pH in particular. The ion exchange was characterized by XRF on solid samples and ion chromatography on liquid samples at higher concentrations and using radioactive 85Sr with scintillation counters to achieve the greater sensitivity needed for low concentration experiments. HIPing and subsequent leach testing of AV-7 and AV-3 was performed in attempts to produce dense ceramic phases suitable for wasteforms, however, neither AV-7 nor AV-3 appeared to form phases suitable for wasteforms due to high leach rates and poor mechanical properties.

158

Chapter 6: Summary

6.1 Conclusions

During this work the main focus has been to identify and characterize potential new materials for strontium ion exchange from nuclear waste streams or ponds. A number of brief surveys into possible routes for producing new materials were investigated with one new, as yet unidentified, material produced. Sn-kostyelvite (AV-7), petarasite (AV-3) and Ca-titanite were hydrothermally synthesised and the ion exchange properties explored with reference to a number of other known ion exchange materials. Strontium ion exchanged AV-7, AV-3 and

Ca-titanite thermal decomposition products were investigated as were the phases formed during hot isostatic pressing of AV-7 and AV-3. Whilst AV-3 and AV-7 both demonstrate overall good ion exchange properties, with AV-3 exhibiting superior uptake to AV-7 over the low concentration range, neither currently perform substantially better than the currently established technology under the conditions examined.

6.1.1 Titanosilicates and zirconosilicates

Three porous materials, fresnoite, titanite and petarasite, have been synthesised hydrothermally. Whereas petarasite has been shown to undergo ion exchange, titanite and fresonite do not appear to do so. Upon ion exchange bond valence sums indicate the most favourable location to be on the Na1 site as at this site the valence sum, 2.03, is approximately equal to 2 for the Sr2+ cation. Upon heating, petarasite decomposes into a parakeldashite phase and another unidentified phase. Which phase the strontium is located in

159 could not be established and further work would be necessary before the suitability of this phase as a waste form can be determined.

Titanite and fresnoite, however, do not appear to ion exchange and upon heating the structures appear to be stable up to 1000 °C. Whilst all of these materials have porous structures of 5 or 6 rings, only petarasite contains water molecules within the pores which may play an important role in the ion exchange process.

6.1.2 Stannosilicates

A brief investigation into the synthesis of a new stannosilicate phase for ion exchange was performed. Initial work suggests that due to the stability of SnO2 aggressive conditions from high concentrations of hydroxide or fluoride ions are necessary for the crystallization of these types of materials.

A new sodium stannosilicate has been prepared using SnII oxalate, with a cubic structure which undergoes a phase transition to a monoclinic phase upon cooling to 100 K. Structure solution is still on-going but, having obtained high-quality synchrotron PXRD data along with solid state Si and Sn NMR, ICP/MS for elemental composition and TGA data, solution should be possible given time.

AV-7, a synthetic Sn form of kostylvite has also been prepared with good quality TOF NPD obtained to investigate the structure changes which occur upon strontium ion exchange with comparison to the reported structure in literature. The thermal decomposition product of Sr- loaded AV-7 has also been prepared and the structural changes investigated and compared to structural data published in literature.

160

6.1.3 Ion exchange

A variety of methods were used to characterize the ion exchange properties of AV-7, AV-3 and titanite with comparisons to Nb-CST and clinoptilolite. XRF spectrophotometry was used to characterize solid samples, however, much of the acquired data proved to be inconsistent and required normalizing to obtain sensible information. Frequently problems also arose with sample preparation causing fused beads to split and crack. XRF is also unsuitable for measuring aqueous samples due to X-rays penetrating through the sample which causes errors with the measurement as the backing plate of the sample holder is detected.

Liquid samples were measured with ion chromatography, which had a high throughput and reasonable sensitivity and measures multiple components in the solution so sodium or potassium exchanged from the solids into the solution can be quantified in addition to strontium. This technique was used to look at some ion exchange experiments in the 100-200 ppm region of concentrations, comparing AV-7, AV-3 and titanite with clinoptilolite and examining the competitive ion exchange of AV-3 under acidic conditions. However, the technique lacked the precision and sensitivity necessary for extremely low level exchanges which necessitated the use of 85Sr and a scintillation counter.

85Sr ion exchanges were performed in three batches, the first batch of experiments examined the ion exchange of a variety of ion exchange materials, AV-3, AV-7, Nb-CST and clinoptilolite under slightly acidic conditions ~pH 6. These initial exchanges indicated very low uptake for titanite which when examined with data obtained from ion chromatography, is likely due to surface adsorption rather than ion exchange. The second batch of 85Sr ion

161 exchanges were performed at much lower concentrations, < 1 ppb, where the whole concentration of strontium comes only from the added 85Sr. These exchanges were performed over neutral (~pH 7) and acidic (~pH 4) conditions with and without competing cations.

These experiments indicated very poor performance of AV-7 vs clinoptilolite and Nb-CST under these conditions, but AV-3 appeared to perform well. However the differences in measured activity are minimal and the solid samples which would have a much lower error could not be measured due to differences in the geometry of the instrument. Also at this concentration, it is possible that the process is dominated by surface adsorption rather than ion exchange so the material with the highest surface area would appear to have the greatest strontium uptake.

The final batch of 85Sr ion exchanges improved upon the methods used previously, a new detector was purchased to enable a change in instrument setup allowing for more lead shielding to be used to reduce background radiation and improve the accuracy of the measurements. The exchange solutions used in this case were pH 10-11 in contrast to the acidic solutions used previously, also these exchange experiments had a number of competing cations at varying concentrations across the four solutions, the concentrations based upon the advice of contacts from the NNL. Under these conditions AV-7 again performed poorly in comparison to the other materials investigated, whereas AV-3 showed higher uptake than clinoptilolite but inferior to Nb-CST in most cases. Ion exchanges performed using solutions with higher concentrations of sodium, caesium or magnesium do not appear to affect significantly the strontium uptake of AV-3. Nb-CST shows some anomalous results in solutions C and D, which have high concentrations of magnesium for solution C and sodium and potassium for solution D, as a significantly greater uptake for

162 these solutions is shown in comparison to solution A which does not contain high concentrations of any competing cations, although it does have a lower concentration of strontium. A general trend for the other materials is visible in that whilst magnesium and calcium do not appear to have a great effect at high pH, high concentrations of sodium and potassium are clearly detrimental to strontium uptake of these materials. Overall for the low concentration ion exchange experiments AV-7 shows poor strontium uptake and selectivity in comparison to other tested materials, whereas AV-3 appears to perform well in comparison to clinoptilolite but inferior to Nb-CST. Higher concentration ion exchange experiments indicate both AV-7 and AV-3 show poorer strontium uptake than clinoptilolite with AV-3 performing better than AV-7. Maximum strontium uptake on these materials may be dependent on differing factors depending on the concentrations of strontium and the concentrations of competing cations. If low concentration strontium uptake is dependent upon surface adsorption then the higher uptake of AV-3 at low concentrations could be explained by what appears to be a high surface area due to the morphology of the particles, although surface area measurements would be necessary to corroborate this.

Hot isostatic pressing was performed on strontium exchanged AV-7 and AV-3 to form dense phases suitable for leach testing, however, PXRD of the produced materials indicate that AV-

7 appears to have amorphized and only one phase from the HIPed AV-3 has been identified.

Due to problems with the HIP can, only 4 samples of each suitable for leach testing were able to be prepared. Following the procedure for MCC(C 1220) 3 repeats are necessary for 30 day leaching samples leaving one sample left for 14 days. After acquiring ICP-MS/OES elemental analysis normalized elemental mass loss values were calculated which revealed high leach rates for both AV-7 and AV-3, which render these materials along with these particular hipping conditions used unsuitable for producing good wasteform materials.

163

6.2 Further work

There are a number of areas which could be expanded upon in this work. In terms of synthesis work the synthesis of noonkanabahite or related phases batisite and scheherbakovite was not achieved. Whilst these are not common minerals, they do occur in nature and it should be possible to prepare them using hydrothermal methods, although long crystallization times may be necessary. The use of a cold-seal vessel, also known as a Tuttle bomb could potentially be of use in this area to reach higher temperatures and pressures, up to 1000°C and 1500 bar, not possible with a standard Parr acid digestion vessel. In general much of the synthesis work has been too broad in scope and could have benefited from more focus and a more specific series of experiments designed to more fully probe the synthesis conditions necessary to form stannosilicate materials. The structure of what appears to be a new stannosilicate material as discussed in Chapter 4 needs more work to solve the structure; with the data already acquired it should be possible to contruct a plausible model of the material given time.

With regards to ion exchange work, explaining why fresnoite and titanite appear to be poor ion exchange materials could be aided using computer modelling to calculate the relative stability of the different cation-exchanged forms and examining how unfavorable the creation of cation vacancy defects are as, in order for an ion exchange reaction to occur, cations must be able to be removed from the cation site. Having found adequate methods of measuring the ion exchange properties of these materials using XRF and 85Sr studies, more detailed work could be performed to produce ion exchange isotherms detailing the change in behaviour more thoroughly. The ion exchange over a range of pH conditions should be investigated over a wider range with more data points to allow for a more definitive trend to be observed; similarly the effect of introducing competing cations should be investigated using a range of concentrations of each of these cations also over a range of pH conditions to reveal how the

164 change in pH and thus the hydration sphere of all of the species in solution affect the ion exchange in terms of kinetics and overall uptake.

Also whilst some surface area and particle size measurements were attempted on AV-7 and

AV-3 to help explain ion exchange trends at low concentraions, the obtained data was not of sufficient quality, as it produced nonsensical figures. This may be due to nitrogen not being the correct gas adsobant to use for these systems. Using porosimetry instruments to acquire surface areas and good isotherms of gas adsorption and comparing these to isotherms produced from ion exchange kinetics studies may aid in resolving the mechanisms of ion exchange or surface adsorption occuring at various concentrations.

The HIPing and leach testing experiments did not provide a great deal of useful information due to the degredation of the samples and limited number of samples which necessitated a smaller time scale. Exploring the different conditions possible when HIPing, and different sample preparation, would allow for better leach testing of the densified forms of strontium exchanged AV-7 and AV-3

165

References

1 DECC and NDA, Radioactive Wastes in the UK : A Summary of the 2013 Inventory, 2013. 2 D. Delacroix, J. P. Guerre, P. Leblanc and C. Hickman, Radionuclide and Radiation Protection Data Handbook 2002, Nuclear Technology Publishing, 2nd edn., 2002. 3 H. S. Thompson, J. R. Agric. Soc., 1850, 11, 313–379. 4 1954. 5 A. Clearfield, Inorganic Ion exchange materials, CRC Press, 1982. 6 A. Clearfield and J. A. Stynes, J. Inorg. Nucl. Chem., 1964, 26, 117–129. 7 L. Jukka, Encyclopedia of Separation Science III, The Nuclear Industry: Ion Exchange, Academic Press, 2000. 8 R. Harjula, A. Paajanen, J. Lehto, E. Tusa and P. Standring, in Proc. Waste Management, Tuscon, Arizona, 2003. 9 K. R. Rao, Main, 2001, 81, 1534–1546. 10 E. Tusa, in Proc. Waste Management, Tuscon, Arizona, 2001. 11 D. R. Lewis, Clays Clay Miner., 1952, 1, 54–69. 12 C. Ma and R. A. Eggleton, Clays Clay Miner., 1999, 47, 174–180. 13 Y. Cho and S. Komarneni, Appl. Clay Sci., 2009, 44, 15–20. 14 S. W. Bailey, Reviews in Mineralogy vol 13: Micas, Bookcrafters Inc, Chelsea, Michigan, 1981. 15 R. Fricke, H. Kosslick, G. Lischke and M. Richter, Chem. Rev., 2000, 100, 2303–2405. 16 J. D. Sherman, Proc. Natl. Acad. Sci., 1999, 96, 3471–3478. 17 M. Howden and j Pilot, Ion Exchange Technology, Elis Horwood, London, 1984. 18 D. M. . Horsley and M. Howden, Inst. Chem. Eng., 2005, 68b, 141. 19 G. Alberti, U. Costantino and M. Pelliccioni, J. Inorg. Nucl. Chem., 1973, 35, 1327– 1338. 20 V. Veselý, Talanta, 1972, 19, 219–262. 21 G. Alberti, Acc. Chem. Res., 1978, 11, 163–170. 22 A. Clearfield, Solvent Extr. Ion Exch., 2000, 18, 655–678. 23 L. H. Baetslé, D. Van Deyck and D. Huys, J. Inorg. Nucl. Chem., 1965, 27, 683–695. 24 A. Dyer, J. Newton and M. Pillinger, Microporous Mesoporous Mater., 2009, 126, 201–212. 25 A. Dyer, R. Harjula, J. Newton and M. Pillinger, Microporous Mesoporous Mater., 2010, 130, 63–66. 26 A. Dyer, R. Harjula, J. Newton and M. Pillinger, Microporous Mesoporous Mater., 2010, 135, 21–29. 27 A. Dyer, J. Newton and M. Pillinger, Microporous Mesoporous Mater., 2010, 130, 56– 62.

166

28 A. Dyer, J. Newton and M. Pillinger, Microporous Mesoporous Mater., 2009, 126, 192–200. 29 L. Jukka, L. Brodkin and R. Harjula, Radioact. Waste Manag. Enviromental Remediat. - ASME 1997, 1997, 245–248. 30 D. T. Hobbs, M. J. Barnes, R. L. Pulmano, K. M. Marshall, T. B. Edwards, M. G. Bronikowski and S. D. Fink, Sep. Sci. Technol., 2005, 40, 3093–3111. 31 T. B. Peters, M. J. Barnes, D. T. Hobbs, D. D. Walker, F. F. Fondeur, M. A. Norato, S. D. Fink and R. L. Pulmano, Sep. Sci. Technol., 2006, 41, 2409–2427. 32 A. Clearfield, A. I. Bortun, L. N. Bortun, D. M. Poojary and S. A. Khainakov, J. Mol. Struct., 1998, 470, 207–213. 33 S. Solbrå, N. Allison, S. Waite, S. V Mikhalovsky, A. I. Bortun, L. N. Bortun and A. Clearfield, Environ. Sci. Technol., 2001, 35, 626–629. 34 A. M. Puziy and J. R. García, J. Radioanal. Nucl. Chem., 1999, 240, 851–857. 35 N. Talaie, H. R. Aghabozorg and S. Alamdar Milani, J. Radioanal. Nucl. Chem., 2012, 292, 473–479. 36 D. M. Poojary, R. a. Cahill and A. Clearfield, Chem. Mater., 1994, 6, 2364–2368. 37 A. Bortun, L. Bortun and A. Clearfield, Abstr. Pap. Am. Chem. Soc., 1997, 213, 133– IEC. 38 US4853202 A, 1987. 39 J. Rocha, P. Brandão, M. W. Anderson, T. Ohsuna and O. Terasaki, Chem. Commun., 1998, 667–668. 40 L. Liu, R. Singh, G. Li, P. Xiao, P. Webley and Y. Zhai, J. Hazard. Mater., 2011, 195, 340–5. 41 C. C. Pavel, K. Popa, N. Bilba, a. Cecal, D. Cozma and a. Pui, J. Radioanal. Nucl. Chem., 2003, 258, 243–248. 42 L. Lv, G. Tsoi and X. S. Zhao, Ind. Eng. Chem. Res., 2004, 43, 7900–7906. 43 A. Nalaparaju, Z. Q. Hu, X. S. Zhao and J. W. Jiang, J. Memb. Sci., 2009, 335, 89–95. 44 C. C. Pavel, M. Walter, P. Pöml, D. Bouëxière and K. Popa, J. Mater. Chem., 2011, 21, 3831. 45 N. J. Coleman, S. P. Lewis, A. P. Mendham and V. Trivedi, J. Porous Mater., 2010, 17, 747–753. 46 V. Sebastian, I. Diaz, C. Tellez, J. Coronas and J. Santamaria, Adv. Funct. Mater., 2008, 18, 1314–1320. 47 C. S. Fewox and A. Clearfield, J. Phys. Chem. A, 2008, 112, 2589–2597. 48 C. S. Fewox, S. R. Kirumakki and A. Clearfield, Chem. Mater., 2007, 19, 384–392. 49 C. S. Fewox, A. Clear and A. J. Celestian, 2011, 3596–3604. 50 R. G. Anthony, R. G. Dosch, D. Gu and C. V Philip, Ind. Eng. Chem. Res., 1994, 33, 2702–2705. 51 L. Al-Attar, A. Dyer and R. Harjula, J. Mater. Chem., 2003, 13, 2963. 52 H. Xu, A. Navrotsky, M. Nyman and T. M. Nenoff, J. Am. Ceram. Soc., 2005, 88, 1819–1825.

167

53 S. Chitra, S. Viswanathan, S. V. S. Rao and P. K. Sinha, J. Radioanal. Nucl. Chem., 2010, 287, 955–960. 54 S. Chitra, R. Sudha, S. Kalavathi, a. G. S. Mani, S. V. S. Rao and P. K. Sinha, J. Radioanal. Nucl. Chem., 2012. 55 D. V Marinin and G. N. Brown, Waste Manag., 2000, 20, 545–553. 56 Y. a. Uvarova, E. Sokolova, F. C. Hawthorne, R. P. Liferovich and R. H. Mitchell, Can. Mineral., 2003, 41, 1193–1201. 57 K. O. H. Zro, G. D. Ilyushin and L. N. Dem, Inorg. Mater., 2002, 38, 612–616. 58 S. Jale, A. Ojo and F. Fitch, Chem. Commun., 1999, 411–412. 59 L. N. Demianets, Prog. Cryst. Growth Charact. Mater., 1991, 21, 299–355. 60 Z. Lin, J. Rocha, J. D. Pedrosa de Jesus and A. Ferreira, J. Mater. Chem., 2000, 10, 1353–1356. 61 R. P. Nikolova, K. Fujiwara, N. Nakayama and V. Kostov-Kytin, Solid State Sci., 2009, 11, 382–388. 62 R. P. Nikolova, B. L. Shivachev and S. Ferdov, Microporous Mesoporous Mater., 2013, 165, 121–126. 63 N. R. Petrova, N. Noriaki, B. Snejana, B. Petr and K.-K. Vladislav, Solid State Sci., 2011, 13, 1187–1190. 64 A. Ferreira, Z. Lin, M. R. Soares and J. Rocha, J. Solid State Chem., 2010, 183, 3067– 3072. 65 Z. Lin, J. Rocha, P. Ferreira, A. Thursfield, J. R. Agger and M. W. Anderson, J. Phys. Chem. B, 1999, 103, 957–963. 66 J. Rocha, P. Ferreira, Z. Lin, J. R. Agger and M. W. Anderson, Chem. Commun., 1998, 1269–1270. 67 Z. Lin and J. Rocha, Microporous Mesoporous Mater., 2006, 94, 173–178. 68 Z. Lin and J. Rocha, Stud. Surf. Sci. Catal., 2002, 319–325. 69 Z. Lin and J. Rocha, Microporous Mesoporous Mater., 2004, 76, 99–104. 70 Y. Xie, F. Chang, C. Li, J. Chen, J. Luo, L. Li and X. Hu, CLEAN - Soil, Air, Water, 2014, 42, 521–527. 71 S. S. Mali, C. A. Betty, P. N. Bhosale, R. S. Devan, Y.-R. Ma, S. S. Kolekar and P. S. Patil, CrystEngComm, 2012, 14, 1920. 72 R. Jenkins and R. L. Snyder, Introduction to X-Ray Powder Diffractometry, Wiley- Interscience, 1996. 73 W. H. Bragg and W. L. Bragg, Proc. R. Soc. A Math. Phys. Eng. Sci., 1913, 88, 428– 438. 74 C. Hammond, The Basics of Crystallography and Diffraction, International Union of Crystallography Oxford Science Publications, Third Edit., 2009. 75 P. P. Ewald, Acta Crystallogr. Sect. A Cryst. Physics, Diffraction, Theor. Gen. Crystallogr., 1969, 25, 103–108. 76 A. Guinier, X-ray diffraction in crystals, imperfect crystals and amorphous bodies, W.H. Freeman, 1963.

168

77 C. Baerlocher, A. K. Cheetham, W. I. F. David, T. H. de Keijser, R. Delhez, R. J. Hill, F. Izumi, J. D. Jorgensen, J. I. Langford, D. Louer, E. J. Mittemeijer, E. Prince, J. W. Richardson Jr, H. M. Rietveld, T. M. Sabine, E. J. Sonneveld, R. L. Snyder, P. Suortti, H. Toraya, R. B. Von Dreele and R. A. Young, The Rietveld Method, 1995. 78 B. Beckhoff, B. Kanngieber, N. Langhoff and R. Wedell, Practical X-Ray Fluroescence Analysis, Springer. 79 H. Bennet and G. Oliver, XRF Analysis of Ceramics Minerals and Allied Materials, John Wiley & Sons, 1992. 80 Y. A. Uvarova, E. Sokolova, F. C. Hawthorne, R. P. Liferovich, R. H. Mitchell, I. V. Pekov and A. E. Zadov, Mineral. Mag., 2010, 74, 441–450. 81 D. M. Poojary, a I. Bortun, L. N. Bortun and A. Clearfield, J. Solid State Chem., 1997, 132, 213–223. 82 J. Rocha, P. Ferreira, Z. Lin, J. R. Agger, M. W. Anderson, R. A. J. and W. A. M., Chem.Commun., 1998, 1269–1270. 83 D. G. Medvedev, A. Tripathi, A. Clearfield and J. Hanson, 2004, 3659–3666. 84 R. Hall and J. Readman, in Acta Cryst., 2014, p. C1526. 85 V. I. Starichenko, Russ. J. Ecol., 2011, 42, 64–70. 86 G. . Gibbs and J. A. Speer, Am. Mineral., 1976, 61, 238–247. 87 M. Taylor and G. E. Brown, Am. Mineral., 1976, 61, 435–447. 88 P. B. Moore and J. Louisnathan, Science (80-. )., 1967, 156, 1361–1362. 89 5110568, 1992. 90 Z. Lin, A. Ferreira and J. Rocha, J. Solid State Chem., 2003, 175, 258–263. 91 D. Wu, C. Han, S. Wang, N. Wu and I. A. Rusakova, Mater. Lett., 2002, 53, 155–159. 92 W. Clegg, X-Ray Crystallography, Oxford University Press, 2nd edn., 2015. 93 J. Evans and Bill David, .

169